WO2005029109A1

WO2005029109A1 - Noise filter for three-dimensional heterogeneous nuclear correlation nmr spectrum, noise filter method, and noise filter program

Info

Publication number: WO2005029109A1
Application number: PCT/JP2004/013811
Authority: WO
Inventors: Shigeyuki Yokoyama; Takanori Kigawa; Naohiro Kobayashi
Original assignee: Riken
Priority date: 2003-09-22
Filing date: 2004-09-22
Publication date: 2005-03-31
Also published as: JP4381077B2; JP2005098753A

Abstract

A noise filter for easily removing noise from a three-dimensional heterogeneous nuclear correlation spectrum in a short time without need to relegate a signal, a noise filter method, and a noise filter program are disclosed. From a measured three-dimensional heterogeneous nuclear correlation NMR spectrum, peak groups G (m) (m is a natural number) each including a predetermined number or more of peaks observed to be in line along the y-axis are detected. A mask file in which the reference x-z plane coordinates of each peak group G (m) are recorded as the coordinates m of a mask peak is prepared. The reference x-z plane coordinates of each measured peak are compared with the coordinates m of the mask peak. The peak with no corresponding mask peak is judged as a noise and removed from the three-dimensional heterogeneous nuclear correlation NMR spectrum.

Description

Specification

Noise filter device, noise filter method, and noise filter program for three-dimensional heteronuclear correlation NMR spectrum

Technical field

The present invention relates to a noise filter device, a noise filter method, and a noise filter program for a three-dimensional heteronuclear correlation NMR spectrum.

Background art

[0002] With the rapid development of genomic informatics in the last few years, NMR stereostructure analysis of proteins has been strongly demanded for both speed and accuracy! Puru.

[0003] In Nuclear Magnetic Resonance (NMR), the nuclear Overhauser effect ( _NOE ) observed between proton nuclei existing within 5A is indispensable structural information for a method of analyzing the three-dimensional structure of a protein. NOEs generated between proton nuclei at the middle or distal position in the amino acid sequence have middle-range NOEs! /, Are called long-range NOEs, and the distance information between proton nuclei obtained by these NOEs is It is used as a distance limit in calculations, and the number strongly affects the accuracy and reliability of the calculation structure. At present, the number of observable NOEs has increased remarkably due to the progress of NMR measurement methods, and a higher-order structure calculation method from NOEs has been proposed (for example, see Non-Patent Document 1).

[0004] Non-patent document l: Guntert, P. et al, J. Mol. Biol., 273, 283-298 (1 997)

Disclosure of the invention

Problems to be solved by the invention

[0005] Under such circumstances, in recent years, a fully automatic assignment method of NOE such as a program CANDID has been announced, and it is beginning to be expected that three-dimensional structure calculations by NMR can be performed quickly. In actual NOE automatic homing method, a chemical shift table created by attributable to the total signal observable user, from 'H-' C HSQC- NOESY space Tuttle and ¹ H- ¹⁵ N HSQC- NOESY spectrum The user can manually enter the NOE by entering a chemical shift table (NOE table) with semi-automatically detected peaks. The automatic attribution method is achieved. Therefore, how accurate the NOE table is created is an important point of high-precision three-dimensional structure calculation.

[0006] However, actually measured by ¹ H- ¹³ C HSQC-NOESY obtained, ^ - 'N HS QC- NOESY removal operation of the noise generated in the spectrum, which still bothering analyst However, depending on the quality of the sample and the quality of the spectrum, removing noise alone is a great deal of labor and time.For example, when trying to analyze the three-dimensional structure of a protein with about 100-150 residues, It took 2-3 days to remove the noise.

[0007] In the prior art, after calculating the three-dimensional structure with a certain degree of accuracy, the NOE peak list that was detected manually was used to calculate the expected three-dimensional structure NOE, and the NOE that could not be detected due to the overlap of the peaks was calculated. The method of adding is already shown. However, no method has been established to remove the noise before the signal is attributed.

[0008] The present invention has been made in view of the above, and requires signal assignment in a three-dimensional heteronuclear correlation NMR ^ vector, which is one of the powerful measurement means in NMR three-dimensional structure analysis of proteins. It is an object of the present invention to provide a noise filter device, a noise filter method, and a noise filter program that can remove noise easily and in a short time.

Means for solving the problem

[0009] To achieve the above object, the invention according to claim 1, X-axis, different other than ¹ H nuclei obtained by time expansion corresponding to direct observation to have ¹ H nuclei of chemical shift values C Nuclear chemistry shift

H

Triaxial force on the z-axis corresponding to the default value C and the y-axis corresponding to other parameters.

X

A noise filter device for removing three-dimensional heteronuclear correlation NMR spectrum color noise, wherein in the three-dimensional heteronuclear correlation NMR spectrum, a predetermined number or more of peaks are substantially linear on the y-axis. The X-z plane reference coordinates (P (m), P (m)) of the peak group G (m) (m is a natural number) observed alongside are defined as the coordinates of the mask peak m.

H X

A mask file in which the coordinates of the peak m are recorded in advance, and for each peak n (n is a natural number) in the three-dimensional heteronuclear correlation N MR spectrum, for each peak n, the X-z plane reference coordinates (C ( n), C (n))

H X

For each peak n in the peak table, the X-z plane reference coordinates of the peak n ( C (n), C (n)) and the coordinates (P (m), P (m)) of each mask peak m in the mask file

H X H X

Is calculated (C (n) —P (m), C (n) —P (m)), and the coordinate difference is within a predetermined range.

H H X X

And a corresponding mask peak searching means for searching for a mask peak included in the corresponding mask peak as a corresponding mask peak of this peak n. And a first noise removing means for removing the peak table force stored in the storage means.

[0010] The invention according to claim 2 is the noise filter device according to claim 1, further comprising a peak table creating unit that creates the peak table.

[0011] Further, the invention according to claim 3 is characterized in that the noise filter device according to claim 1 or 2 further comprises a mask file creating means for creating the mask file.

[0012] Further, in the invention according to claim 4, in the noise filter device according to any one of claims 13 to 13, the predetermined range is set corresponding to a half width of each peak n. It is characterized by

[0013] Further, according to the invention according to claim 5, in the noise filter device according to any one of claims ^{14 to 14} , the z-axis corresponds to a chemical shift value of a 15N nucleus or a ^13C nucleus. The feature is that it is the axis that performs.

[0014] Further, the invention according to claim 6 is the noise filter device according to any one of claims 115, wherein the three-dimensional heteronuclear correlation NMR spectrum is a three-dimensional HSQC-NO ESY spectrum. And the y-axis represents the chemical shift value C of the ¹ H nucleus obtained by time expansion.

HC

It is characterized by corresponding.

According to a seventh aspect of the present invention, there is provided the noise filter device according to any one of the sixteenth to sixteenth aspects, wherein a predetermined number of peaks are sequentially set in order of the peak table force and the peak intensity. Are extracted as upper peaks, and each of the extracted upper peaks is located at coordinates substantially symmetric with respect to any one of the three axes of the X-axis, y-axis, and z-axis with the upper peak as a center. And a peak pair detecting means for detecting a peak pair of a pair of peaks having substantially the same peak intensity, and discriminating peaks constituting the peak pair as noise. A second noise removing unit for removing information of the determined peak from the peak table stored in the storage unit.

[0016] Further, the invention according to claim 8 provides the noise filter device according to any one of claims 117 to 17, having the X coordinate and the y coordinate having substantially the same value from the peak table. All the peaks are extracted as diagonal peaks, and for each extracted diagonal peak, the same sign is applied from the diagonal peak to any one of the three axes of the X axis, y axis, and z axis. A continuous peak group detecting means for detecting a peak group having continuous peaks as a continuous peak group, and determining peaks other than diagonal peaks included in the continuous peak group as noise and determining peaks determined as noise. And a third noise removing means for removing information from the peak table stored in the storage means.

[0017] The invention according to claim 9 is the noise filter device according to any one of claims 5 to 8, wherein the peak table having substantially the same value of the X coordinate and the y coordinate is obtained from the peak table. Are extracted as diagonal peaks, and for each of the extracted diagonal peaks, a group of non-diagonal peaks other than the diagonal peaks in which a predetermined number or more peaks are arranged substantially linearly from the diagonal peak in the y-axis direction. A non-diagonal peak group detecting means for detecting as a peak group, and for each detected non-diagonal peak group, determining which of the positive or negative peaks the non-diagonal peak group contains, A code determining means for determining a code as a code of the off-diagonal peak group; a code determined for the off-diagonal peak group by the code determining means; and a code of each peak included in the off-diagonal peak group. Compare with , Characterized in that a fourth noise removal means for the Pikute one Bull force removes a peak with a different sign and determined sign for the non-diagonal peaks as noise.

[0018] Further, the invention according to claim 10 is the noise filter device according to any one of claims 11 to 19, wherein the fifth noise removing means removes a peak corresponding to noise derived from light water from the peak table. It is characterized by having a removing means.

The invention according to claim 11 corresponds to the chemical shift value C of the ¹ H nucleus directly observed.

H

X axis, chemical shift value of heteronucleus other than ¹ H nucleus obtained by time expansion z corresponding to C

X

A three-dimensional heteronuclear correlation NMR spectrum force corresponding to the y-axis corresponding to the y-axis and other parameters NMR spectral force A noise filter method for removing noise, wherein the actually measured three-dimensional From the heteronuclear correlation NMR spectrum, peaks having a predetermined peak intensity or more are extracted as peak n (n is a natural number), and for each extracted peak, the X-z plane reference coordinates (C (n),

H

C (n)) and a peak table creation step to create a peak table

X

A peak group G (in which a predetermined number or more of peaks are observed in a substantially straight line in the axial direction of the y-axis in the storage step of storing the peak table and in the three-dimensional heterogeneous nuclear correlation vector, m) (m is a natural number), the x-z plane reference coordinates (P (m), P (m)) of each peak group G (m) are determined, and each determined X-z plane Reference coordinates (P (m),

H X H

P (m)) to create a mask file that records the coordinates of the mask peak m.

X

And Le creating step, for each peak _n of the peak table of the peak n X- z Rights plane reference coordinates (C (n), C ( n)) and the coordinate of each mask peak m (P (m), P (m))

H X H X

Calculate the coordinate difference (C (n) — P (m), C (n) — P (m)) and calculate from the mask file

H H X X

In the corresponding mask peak search step of searching for a mask peak whose coordinate difference falls within a predetermined range as the corresponding mask peak of this peak n, and in the corresponding mask peak search step, there is an effort to search for a corresponding mask peak. A first noise removing step of determining a peak as noise and removing information of the peak determined as noise from the peak table stored in the storing step.

The invention according to claim 12 corresponds to the chemical shift value C of the ¹ H nucleus directly observed.

H

X

A program for causing a computer to execute a noise filter method for removing three-dimensional heteronuclear correlation NMR spectrum noise from a three-axis force on the y-axis corresponding to the y-axis and other parameters. From the three-dimensional heteronuclear correlation NMR spectrum, peaks having a predetermined peak intensity or higher are extracted as peak n (n is a natural number), and for each extracted peak, the X—z plane reference coordinates (C (n), C (n) Create a peak table corresponding to)

H X

A peak table creating step, a storing step of storing the peak table, and in the three-dimensional heteronuclear correlation NMR spectrum, a predetermined number or more of peaks are arranged substantially linearly in the y-axis direction. For the observed peak group G (m) (m is a natural number), the X-z plane reference coordinates (P (m), P (m)) of each peak group G (m) are determined and determined.

H X

X-z plane reference coordinates (P (m), P (m)) recorded as the coordinates of the mask peak m A mask file creating step of creating a mask file, and for each peak n of the peak table, the X-z plane reference coordinates (C (n), C (n)) of the peak n and the mask

H X

Calculate the coordinate difference (C (n) —P (m), C (n) —P (m)) from the coordinate (P (m), P (m)) of peak m.

H X H H X X

A corresponding mask peak search step of searching for a mask peak whose coordinate difference falls within a predetermined range from a mask file of the mask file as a corresponding mask peak of this peak n; and A first noise removal step of determining a peak for which a corresponding mask peak could not be retrieved as noise and removing information of the peak determined to be noise from the peak table stored in the storage step. Is executed.

The invention's effect

[0021] The noise filter device according to the present invention (claim 1) is capable of filtering a spurious signal called off-signal noise existing in a region different from the peak group G (m) arranged substantially linearly in the y-axis direction. This is to detect and remove the signal. If the off-signal noise can be removed efficiently and accurately before the signal is assigned, the effect is obtained.

[0022] Further, the noise filter device according to the present invention (claim 2) includes a peak table creating means for automatically creating a peak table from actually measured spectrum data. The time and labor required to create a search peak table can be saved, and noise can be more efficiently removed.

In addition, since the noise filter device according to the present invention (claim 3) includes a mask file creating means for automatically creating a mask file from actually measured spectrum data, a user can manually create a peak group. The time and effort required to create a search mask file can be saved, and noise can be removed more efficiently.

Further, the noise filter device according to the present invention (claim 4) determines a range corresponding to the half width of each peak n, and searches for a mask peak falling within this range as a corresponding mask peak. Noise can be more accurately removed.

[0025] In addition, the force Cal noise filter device of the present invention (Claim 5), - ¹³ C HSQC-NOE SY spectrum, 'H-' N HSQC-NOESY spectrum, HCCH- COZY spectrum, and, HCCH- TOCSY spectrum Apply to noise removal of spectrum selected from It can be suitably used for NMR three-dimensional structure analysis of proteins.

The noise filter device according to the present invention (claim 6) can be applied to noise removal of three-dimensional HSQC-NOESY spectra, and is suitably used for NMR three-dimensional structure analysis of proteins. it can.

[0027] Further, the noise filter device according to the present invention (claim 7) detects incomplete decoupling noise or wiggle noise observed as a pair around a strong signal such as a methyl signal. This eliminates incomplete decoupling noise and Uyghur noise before assigning the signal.

[0028] Further, the noise filter device according to the present invention (claim 8) provides a tilt noise generated as a distortion of a baseline generated near a diagonal peak derived from a signal having a slow relaxation and a strong intensity such as a methyl signal. Is detected and removed, and it is possible to effectively and accurately remove the till noise before assigning the signal.

[0029] Further, the noise filter device according to the present invention (claim 9) has a non-diagonal peak (NOE peak) in which a predetermined number or more of peaks are arranged substantially linearly in the y-axis direction. Using the fact that the signs (phases) of (h) are theoretically the same, peaks with different signs can be detected as noise and can be removed efficiently and accurately.

[0030] Further, the noise filter device according to the present invention (claim 10) has an effect of removing noise derived from light water.

[0031] The noise filter method according to the present invention (claim 11) is effective in reducing off-signal noise existing in a region different from the peak group G (m) that is substantially linearly aligned with the y-axis direction. If it can be properly and accurately removed, it has an effect.

[0032] Further, the noise filter program according to the present invention (claim 12) is different from the peak group G (m) that is substantially aligned in the y-axis direction by causing the computer to execute the noise filter program. This has the effect of efficiently and accurately removing off-signal noise existing in the region.

Brief Description of Drawings

FIG. 1 is a block diagram of a noise filter device according to a first embodiment. FIG. 2 is an actually measured three-dimensional ¹ H— ¹³ C HSQC—NOESY spectrum diagram.

FIG. 3 is a schematic diagram showing the principle of noise removal performed by the noise filter device according to the first embodiment.

FIG. 4 is a flowchart of noise removal according to the first embodiment.

FIG. 5 is a diagram showing an example of a peak table.

FIG. 6 is a diagram showing an example of a mask finale.

[FIG. 7] FIG. 7 is a flow chart for searching a mask peak corresponding to a color in a mask file to determine noise.

FIG. 8-1 is a partially enlarged view of off-signal noise in FIG. 2.

FIG. 8-2 is a partially enlarged view of the noise filter device according to the first embodiment after removing noise from FIG. 8-1.

FIG. 9 is a block diagram of a noise filter device according to a second embodiment.

FIG. 10 is a flowchart of creating a mask file.

[FIG. 11-1] FIG. 11-1 is a projection view in which reference coordinates of a peak recorded in a peak table are projected on an Xz plane.

[FIG. 11-2] FIG. 11-2 is a projection view showing a section detected as a peak group existence area on the Xz plane in FIG. 11-1.

FIG. 12 is a block diagram of a noise filter device according to a third embodiment.

FIG. 13 is a diagram showing an example of incomplete decoupling noise, in which (A) is a diagram obtained by two-dimensionally slicing in the xy plane, and (B) is a cross-sectional view taken along the line II in (A).

[FIG. 14] FIG. 14 is a diagram showing an example of wiggle noise, in which (A) is a diagram obtained by two-dimensionally slicing in the xy plane, and (B) is a cross-sectional view taken along the line Ι- (A).

[FIG. 15] FIG. 15 is a diagram showing an example of tilt noise, in which (A) a two-dimensional slice on the xy plane, and (B) a cross-sectional view taken along the line (A) of (A).

[FIG. 16] FIG. 16 is a diagram showing an example of different sign noise, in which (A) is a diagram obtained by two-dimensionally slicing in the xy plane, and (B) is a cross-sectional view taken along the line (-(A).

FIG. 17 is a flowchart of noise removal according to the third embodiment.

FIG. 18 is a flowchart of peak pair detection. FIG. 19 is a diagram showing an example of an upper peak table.

FIG. 20 is a flowchart of detecting a continuous peak group.

FIG. 21 is a diagram showing an example of a diagonal peak table.

FIG. 22 is a flowchart of detecting a non-diagonal peak group.

[FIG. 23-1] FIG. 23-1 is a partially enlarged view of the wiggle noise and the incomplete decoupling noise in FIG.

[FIG. 23-2] FIG. 23-2 is a partially enlarged view after noise is removed from FIG. 23-1 by the noise filter device of the third embodiment.

[FIG. 24-1] FIG. 24-1 is a partially enlarged view of the tilt noise in FIG.

[FIG. 24-2] FIG. 24-2 is a partially enlarged view after noise has been removed from FIG. 24-1 by the noise filter device of the third embodiment.

FIG. 25-1 is a partially enlarged view of the different sign noise in FIG. 2.

[FIG. 25-2] FIG. 25-2 is a partially enlarged view after noise is removed from FIG. 25-1 by the noise filter device of the third embodiment.

[FIG. 26-1] FIG. 26-1 is a partially enlarged view of the water noise in FIG.

[FIG. 26-2] FIG. 26-2 is a partially enlarged view after noise is removed from FIG. 26-1 by the noise filter device of the third embodiment.

Explanation of symbols

1 Noise filter device

10 Peak table creation section

11 Mask file creation section

12 Memory

13 Corresponding mask peak search section

14 1st noise canceller

15 Peak pair detector

16 Second noise eliminator

17 Continuous peak group detector

18 Third noise eliminator 19 Off-diagonal peak group detector

20 Sign determination unit

21 Fourth noise remover

22 Water noise detector

23 Fifth noise remover

2 NMR

3 External I / O device

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, preferred embodiments of a noise filter device, a noise filter method, and a noise filter program according to the present invention will be described in detail with reference to the accompanying drawings.

. Note that the present invention is not limited to the embodiments.

(First Embodiment)

FIG. 1 shows a block diagram of a noise filter device according to the first embodiment. The first embodiment is a noise filter device suitable mainly for removing off-signal noise. The noise filter device 1 according to the first embodiment includes a storage unit 12, a corresponding mask peak search unit 13, and a first noise removal unit 14.

[0037] The noise filter device 1 is connected to the NMR 2 and the external input / output device 3. N

The spectrum data measured using the MR 2 is transmitted to the noise filter device 1 and stored in the storage unit 12.

[0038] The external input / output device 3 reads out the spectrum data from the storage unit 12, and displays the three-dimensional heteronuclear correlation NMR spectrum on the screen. The user detects peaks on the screen of the external input / output device 3 while confirming the actually measured 3D heteronuclear correlation NMR spectrum, and for each peak, the 3D reference coordinates, half width, and peak intensity of the peak are detected. Create a peak table corresponding to. The peak table data created by the user is also input to the external input / output device. The data of the peak table input from the external input / output device 3 is transmitted to the storage unit 12 and stored.

[0039] Further, the user detects peak groups observed in a substantially straight line in the axial direction of the y-axis in the actually measured three-dimensional heteronuclear correlation NMR spectrum, and determines the X-z of each peak group G (m). Plane A mask file that records the reference coordinates (P (m), P (m)) as the coordinates of the mask peak m

H X

create. The data of the mask file created by the user is input from the external input / output device 3. The mask file data input from the external input / output device 3 is transmitted to the corresponding mask peak search unit 13.

The corresponding mask peak search unit 13 receives the data of the mask file from the external input / output device 3 and reads the data of the peak table from the storage unit 12. The corresponding mask peak search unit 13 compares the peak table with the mask file, searches the peak table for a peak corresponding to noise, and transmits the peak corresponding to noise to the first noise removal unit 14 as noise data. I do.

The first noise removing unit 14 reads the data of the peak table from the storage unit 12, removes the information of the peak corresponding to the noise from the peak table, and removes the information of the peak corresponding to the noise. The peak table is transmitted to the storage unit 12.

The storage unit 12 stores the data of the peak table after noise removal received from the first noise removal unit 14, and transmits the data to the external input / output device 3 as necessary. The user can check the three-dimensional heteronuclear correlation NMR vector after the noise is removed by the noise filter device 1 on the screen of the external input / output device 3.

Before removing noise using the noise filter device of the present invention, a user measures a three-dimensional heteronuclear correlation NMR spectrum. As used herein, the term “three-dimensional heteronuclear correlation NMR spectrum” refers to a three-dimensional NMR spectrum obtained by combining a two-dimensional heteronuclear correlation spectrum with another parameter. Here, the “two-dimensional heteronuclear correlation spectrum” is obtained by directly measuring the correlation spectrum between ¾ and a heterogeneous nucleus other than ¹ H ( ¹³ C nucleus, ¹⁵ N nucleus, etc.) by ¹ H nuclear observation. , X-axis corresponding to the chemical shift value C of the ¹ H nucleus directly observed), and 直接 Chemical shift value C of the heterogeneous nucleus other than the nucleus

H

And the two axes of the z-axis (heterogeneous nuclear time expansion axis) corresponding to. The present invention can be applied

X

The three-dimensional heteronuclear correlation NMR spectrum is not particularly limited as long as it is a three-dimensional NMR spectrum including the ¹ H nuclear direct observation axis and the heteronuclear time expansion axis.

As a specific example of the three-dimensional heteronuclear correlation NMR ^ vector to which the present invention can be applied, a ¹³ C HSQC—NOESY spectrum (x-axis: nuclear direct observation axis, y-axis: nuclear time expansion) Axis, z axis: "C nuclear time expansion axis", ¹⁵ N HSQC—NOESY spectrum (x axis: direct nuclear observation axis, y axis: ¹ H nuclear time expansion axis, z axis: ¹⁵ N nuclear time expansion axis), HCCH — COSY static (X axis: ¾nuclear direct observation axis, y axis: ¾nuclear time expansion axis, z axis: ¹³ C nuclear time expansion axis), HCC H— TOCSY spectrum (X axis: ¹ H nuclear direct observation axis) Axis, y axis: ¹ H nuclear time expansion axis, z axis: ¹³ C nuclear time expansion axis) and the like.

Hereinafter, the present invention will be described in detail, but it does not prevent application of the present invention to other three-dimensional heteronuclear correlation NMR spectra.

First, the user performs full-automatic peak detection for proteins whose structure has been identified using NMR2. For example, when peak detection is performed by the ¹ H— ¹⁵ N HSQC—NOESY measurement method, the obtained ¹ H— ¹⁵ N HSQC—NOESY spectrum is the HSQC vector between ¹⁵ N (X axis: direct nuclear observation axis, z axis : ¹⁵ N nuclear time expansion axis) and ¹ H NOESY statue (X axis: direct nuclear observation axis, y axis: nuclear time expansion axis).

FIG. 2 shows a three-dimensional ¹ H— ¹³ C HSQC—NOESY vector actually measured for mouse SH3. 3D ¹ H- ¹³ C HSQC- NOESY measurement conditions of the spectrum of FIG. 2, 8 00MHz, mixing time 80 ms, the temperature 25 ° C, sample concentration 1. Omm, 20 mM phosphate buffer over, lOOmM NaCl, 90% HO / 10% DO. Figure 2 is a three-dimensional ¹ H- ¹³ C HSQC

twenty two

This is a two-dimensional slice at ¹³ CZ23.5 ppm of the NOESY spectrum.The horizontal axis is the chemical shift value C of the directly observed ¹ H nucleus, and the vertical axis is obtained by expanding the time.

H

Corresponding to the chemical shift value C of the ¹ H nucleus. The dashed line is on the diagonal of the spectrum

HC

Correspondingly, the peak appearing on this broken line is called the diagonal peak. The NOE to be normally detected is thus observed as a non-diagonal peak linearly distributed in the vertical axis direction from the diagonal peak. In FIG. 2, the detected peaks are shown in blue when they are on the two-dimensional spectrum slice, and the peaks in the upper or lower layer by one slice are shown in green. Thus, the three-dimensional ¹ H— ¹³ C HSQC—NOESY spectrum shows the chemical shift value C of the ¹ H nucleus obtained by time-expanding the X axis corresponding to the chemical shift value C of the ¹ H nucleus directly observed.

H HC

3-dimensional force of the corresponding y-axis and z-axis corresponding to the chemical shift value C of the ¹⁵ N nucleus

N

It can be represented as data. In [0048] 'H-' C HSQC- NOESY spectra, X y plane represents the ^¹ H- ¹ H NOESY's vector, X- z plane represents a HSQC spectrum of ¹³ C. The diagonal peak observed on the xy plane represents the ¹ H chemical shift of the protein. On the other hand, the NOE effect that works only spatially close (shorter linear distance) is observed as a non-diagonal peak (NOE peak). The presence of a NOE peak indicates that they are within about 0.5 nm. Therefore, the secondary and tertiary structures of proteins can be determined by collecting information on the presence or absence of NOE between atoms.

FIG. 3 shows a schematic diagram of the principle of noise removal performed by the noise filter device according to the first embodiment. FIG. 3 (1) is a schematic diagram of the actually measured three-dimensional spectrum, in which peaks derived from signals are shown in black, and peaks corresponding to off-signal noise are shown in white.

[0050] As shown in Fig. 3 (I), in the three-dimensional heteronuclear correlation NMR spectrum, the peak derived from the signal is substantially aligned with the diagonal peak derived from the target sample in the y-axis direction. It is observed as a group of peaks aligned on a straight line. A lot of spurious signals called off-signal noise are usually found in a region different from these peak groups. In order to efficiently remove this off-signal noise, first, in an actually measured three-dimensional heteronuclear correlation NMR spectrum, a predetermined number or more of peaks are observed almost in a line in the y-axis direction. The peak group G (m) (m is a natural number) is detected, and the reference Xz plane coordinates of each peak group G (m) are converted to the coordinates (P (m), P (m))

H N

Create a mask file recorded by Then, the actually measured reference X-Z plane coordinates of each peak are compared with the coordinates (P (m), P (m)) of the mask peak m, and the corresponding mask peak exists.

H N

A peak that does not exist is determined as noise. Then, as shown in FIG. 3 (III), the peak determined as noise is removed from the three-dimensional heteronuclear correlation NMR ^ vector. In the present specification, a noise filter method based on the principle will be described as a mask filter method.

FIG. 4 shows a flowchart for removing noise using the noise filter device according to the first embodiment. Less than,

The HSQC-NOESY spectrum will be described as an example, but this does not preclude the present invention from being applied to other three-dimensional heteronuclear correlation NMR ^ vectors. The threshold value described below for the ¹⁵ N nucleus can be applied to other nuclides such as the ¹³ C nucleus. [0052] First, the user, the ¹ H- ¹⁵ NH SQC- NOESY spectrum was measured for protein to identify the structure using NMR2, acquires the ¹⁵ N HSQC- NOESY spectrum data (Step S 10). ¹ H- ¹⁵ N HSQC- NOESY spectrum data measured by NMR2 is transmitted to the storage unit 12, it is stored.

[0053] The user, based on, creating a peak table were measured ¹ H- ¹⁵ N HSQC- NOESY spectrum (step S 11). To create the peak table, the user reads out the spectrum data from the storage unit 12 through the external input / output device 3 and displays the three-dimensional heteronuclear correlation NMR ^ vector on the screen of the external input / output device 3. The user checks the 'H-'N HSQC-NOESY spectrum on the screen of the external input / output device 3 and detects the peak. n is a natural number). Then, for each peak, a peak table is created by associating the three-dimensional reference coordinates, half width, and peak intensity of the peak.

FIG. 5 shows an example of the peak table. The peak table contains the peak ID, 3D reference coordinates (C (n), C (n), C (n)), half-width in each axis direction, and peak for each peak.

H HC N

It is created corresponding to the strength. C (n) is the X coordinate (chemical shift of the ¹ H nucleus observed directly

H

Value), C (n) is the y coordinate (chemical shift value of ¹ H nucleus obtained by time expansion), C (n) is the z coordinate (

HC N

¹⁵ N nucleus chemical shift value), where w (n), w (n), and w (n) are x-axis, y-axis, and z-axis directions, respectively.

H HC N

And I (n) represents the peak intensity. Here, as three-dimensional reference coordinates

, Peak top coordinates.

The data of the peak table input by the user to the external input / output device 3 is transmitted to the storage unit 12 and stored (step S12).

[0056] Next, the user, ¹ H- ¹⁵ N HSQC- NOESY peaks G (m) to be observed side by side substantially in a straight line in the axial direction of the y axis in the spectrum (m is a natural number) is detected, each peak group

The X-z plane reference coordinates of G (m) (P (m), P (m)) were recorded as the coordinates of the mask peak m

H N

Create a mask file (step S13). The user, through the external input and output device 3, ¹ H- ¹⁵ N HSQC- NOESY confirmed the spectral data, while allowing not adversely picking as much as possible noise visually detects each peak group. The user determines the coordinates that overlap with the largest number of peaks in each peak group visually detected, and calculates these coordinates (P (m), P (m)). Create a mask file that records the coordinates of the peak m. Figure 6 shows an example of a mask file. P (m) represents the X coordinate of the mask peak m, and P (m) represents the z coordinate of the mask peak m

H N

[0057] The mask file data input to the external input / output device 3 by the user is transmitted to the corresponding mask peak search unit 13 (step S13).

[0058] The corresponding mask peak search unit 13 compares the peak table read from the storage unit 12 with the data of the mask file received from the external input / output device 3, and for each peak, stores the corresponding mask in the mask file. It is searched whether a peak exists (step S14). The corresponding mask peak searching unit 13 transmits the peak ID of the peak for which no corresponding mask peak could be searched to the first noise removing unit 14 as noise data (step S15).

FIG. 7 shows a flowchart for searching for a corresponding mask peak from a mask file to determine noise. First, the corresponding mask peak search unit 13 calculates the X—z plane coordinates (C (

H

n), C (n)) in small increments, corresponding to the mask peak m determined by the user.

N

V, the number of peaks n is calculated, and an optimization adjustment value that maximizes the number of peaks corresponding to the mask peak m is determined (step S140). Here, “corresponding to the mask peak m” means that the X—z plane coordinates (C (n), C (n)) of the peak n are the coordinates of the mask peak m (P (m), P (m ) )When

H N H N

It means that they are completely overlapped, that they are powerful, or that they are close enough and far from each other.

Whether the peak n corresponds to the mask peak m is determined by the X—z plane coordinates of the peak n (C (n)

H

, C (n)) and the coordinate (P (m), P (m)) of the mask peak m are within the predetermined range.

N H N

Judgment is made by confirming the power. The evaluation of whether peak n corresponds to mask peak m or not is given by the following function E (n, m), which results in 0 or 1. This function E (n

Η, Ν Η, Ν

, m) means that the peak n with an evaluation value of 1 corresponds to the mask peak m, and the peak n with an evaluation value of 0 means that it corresponds to the mask peak m! / ヽ.

[0061] [Number 1] 0 if | C _H (n) + k _H -P _H (m) |> t _H + w _H (n)

or | C _N (n) + k _N -P _N (m) |> t _N + w _N (n)

E _H , N ( ⁿ , ^m ) ⁼

1 if | C _H (n) + k _H -P _H (m) | <t _H + w _H (n)

and | C _N (n) + k _N -P _N (m) | <t _N + _N (n)

[0062] Here, k and k are fine adjustment values, t and t are threshold values set by the user, and w (n) and w (n) are peak values.

H N H N H N

H is the half-value width of n, corresponding to the X-axis and z-axis, respectively. The threshold values t and t depend on the measurement conditions, etc.

H N

The user can make appropriate settings. For example, to perform an appropriate noise search on spectra measured by NMR at 600--800 MHz, the thresholds t and t should be 0.03 ppm and 0.

H N

It is preferable to set to 3 ppm.

[0063] For optimization, fine adjustment values k, k are changed little by little, and all peaks are set to all mask peaks.

H N

So that the total value of the evaluation values of the results of the evaluations is maximized. That is, the total of the evaluation values is calculated while changing the fine adjustment values k, k at a constant pitch. The pitch width is

H N

Forces that can be determined by the user. For example, in order to perform an appropriate noise search on a spectrum measured by NMR from 600 to 800 MHz, a pitch of Sk = 0.002 in the X-axis direction is usually used.

H

In the z-axis direction, it is preferable to change the fine adjustment value at a pitch of Sk = 0.02.

N

ヽ. That is, k = Sk i, k = Sk j! /, And (Sk = 0.002, Sk = 0.02), i

H H N N H N

By changing the values of and j to 1 50, 1 49, -48 ... 47, 48, 49, 50, respectively, the X z plane coordinates (C (n), C (n)) of peak n 0.lppm— + 0.1

H N

In the ppm and z-axis directions, you can search for a fine adjustment value to optimize the peak n in the range of 1. Oppm-1 "hi. Oppm.

In the optimization, a total value S (i, j) of the evaluation values as a result of evaluating all the peaks with all the mask peaks is calculated, and this value is a maximum, max S (i, j). (I, J).

[0065] [Equation 2]

N M

S (i, j) = ∑ ∑E (n, m)

n = n = Mo

[0066]: The fine adjustment values k = δkI and k = δkJ at the time are optimal for optimizing the peak n. Adjustment value.

Then, the X—z plane coordinates (C (n), C (n)) of the peak n are optimized using the optimization adjustment value.

H N

(Step S141). The X-z plane coordinates of the peak n optimized by the optimization adjustment value are (C (η) + δkI, C (n) + δkJ).

H H N N

Next, the corresponding mask peak search unit 13 uses the optimized x-z plane coordinates (C (η) + δkI, C (n) + δkJ) for each peak n. The corresponding mask in the mask file

H H N N

It is searched whether a peak exists (step S142). Specifically, the corresponding mask peak search unit 13 fixes k = S k I and k = S k J in the function E (n, m), and

Η, Ν Η Η Ν Ν

Determine if there is a mask peak m such that E (n, m) = 1 in the aisle

[0069] [Equation 3]

[0070] If the corresponding mask peak does not exist in the mask file for the peak n! / (Step S142 No), the corresponding mask peak search unit 13 determines the peak ID of the corresponding peak as a noise. The data is transmitted to the first noise removal unit 14 as data (step S143), and the processing for the peak n is completed. If a corresponding mask peak can be detected in the mask file for the peak n (step S 142 Yes), the processing for the peak n ends. As described above, for all the peaks in the peak table, the corresponding mask peak is searched from the mask file to determine whether or not it is noise.

The first noise elimination unit 14 reads the data of the peak table from the storage unit 12 and, based on the noise data received from the corresponding mask peak search unit 13, extracts the information of the peak corresponding to the noise from the peak table. Remove it (Figure 4, step S16). Then, the first noise removing unit 14 transmits a new peak table from which information of the peak corresponding to the noise has been removed to the storage unit 12. FIG. 8-1 is a partially enlarged view of the ¹ H— ¹³ C HSQC—NOESY spectrum of FIG. 2, in which X is assigned to off-signal noise and * is assigned to the NOE peak. FIG. 8-2 is a diagram after the noise of the spectral power of FIG. 8-1 has also been removed by the noise filter device of the first embodiment. Comparing FIGS. 8A and 8B, it can be seen that the off-signal noise marked with X is efficiently removed by the noise filter device of the first embodiment.

In the first embodiment, the force peak table and the mask file in which the X-z plane coordinates of the peak n have been optimized and the force corresponding peak has been searched are correctly created! However, it is not always necessary to optimize, and the corresponding peak may be searched using the Xz plane coordinates described in the direct peak table.

[0074] The noise filter device and the noise filter method according to the first embodiment described above provide a peak group G observed in a substantially straight line with respect to the axial direction of the kernel time expansion axis (y-axis). Noise (off-signal noise) that is observed in a region different from (m) can be efficiently and accurately removed before signal assignment, greatly reducing the burden on analysts involved in noise removal work. .

(Second Embodiment)

The second embodiment is a modification of the first embodiment and is characterized in that a means for automatically creating a peak table and a mask file is provided. FIG. 9 shows a block diagram of a noise filter device according to the second embodiment. In FIG. 9, the same components as those in the first embodiment are denoted by the same reference numerals, and description thereof will be omitted.

The noise filter device according to the second embodiment includes a peak table creation unit 10 and a mask file creation unit 11. The peak table creation unit 10 automatically creates a peak table from the spectrum data stored in the storage unit 12. Further, the mask file creating section 11 automatically creates a mask file using the peak table created by the peak table creating section 10.

[0077] Peak table creation unit 10 creates a ¹ H- ¹⁵ N HSQC- NOESY space Tuttle data force peak table in the following manner. First, the peak table creating unit 10, ^J H - from ¹⁵ N HSQC- NOESY spectrum data, detects the data points corresponding to the peak. ^ — Ν HSQC—NOESY spectral data is Data point (i, j, k) (ijk is an integer, i corresponds to the x coordinate, j corresponds to the y coordinate, and k corresponds to the z coordinate). Can be Peak detection can be performed using a generally used method. Assuming that the data intensity at a data point (i, j, k) of a three-dimensional vector is M (i, j, k), 26 adjacent data points (i 1, j-1, k-1), (i-1, j, k-1), (i-1, j + 1, k-1), (i, j-1, k-1), (i, j, k-- 1), (i, j + 1, k—1), (i + 1, j-1, k—1), (i + 1, j, k—1), (i ++ k—1), (i-1, j-1, k), (i-- k) (i-1, j + 1, k), (i, j--1, k) (i, j + 1, k), (i +1, j-1, k), (i + l, j, k), (i + 1, j + 1, k), (i-1, j-1, k + 1), (i-1 , j, k + l), (i-1, j + 1, k + l), (i, j-1, k + l), (i, j, k + l), (i, j + 1) , k + l), (i + 1, j-1, k + l), (i + 1, j, k + l), (i + 1, j + 1, k + 1) M (g1, j-1, k-1), M (i-1, j, k-1), M (i-1, j + 1, k-1), M (i, j- 1, k—1), M (i, j, k—1), M (i, j + 1, k—1), M (i + 1, j-1, k—1), M (i + 1, j, k—1), M (i + 1, j + 1, k—l), M (i—1, j—1, k) M (i—1, j, k) M (i— 1, j + k) M (i, j-1, k) M (i, j + 1, k), M (i + 1, j-1, k), M (i + 1, j, k) M (i + 1, j + 1, k), M (g1, j-1, k + 1), M (i-1, j, k + 1), M (i-1, j + 1, k + l), M (i, j-1, k + l), M (i, j, k + l), M (i, j + 1, k + l), M (i + 1, j-1, k + l), M (i + 1, j, k + l), M (i + 1, j + 1, k + 1), and those of M (i, j, k) The difference is

[Number 4]

dM (i-l, j-l, kl) = M (i-k) -M (i-l, j-l, k-1) dM (i-l, j, k-l) = M (i , J, k)-M (i-l, j, k 1)

dM (i-l, j + l, k-1 M (i, j, k)-M (i-l, j + l, k-1)

dM (i, j-l, k-l) = (i, j, k) -M (i, j-l, k-l)

dM (i, j, k-l) = M (i, j, k) -M (i, j, k-l)

dM (i, j + l, k-l) = M (i, j, k) -M (i, j + l, k-1)

dM (i + l, l, k-l) = M (i, j, k)-M (i + l, j-l, k-1)

dM (i + l, j, k-l) = M (iJ, k)-(i + l, j, k-l)

dM (i + l, j + l, k-l) = M (i, j, k) -M (i + l, j + l, k-l)

dM (i-l, j-l, k) = (i, j, k) -M (i-l, j-l, k)

dM (i-l, j, k) = M (i, j, k) -M (i-l, j, k)

dM (i-l, j + l, k) = M (i, j, k) -M (i-l, j + l, k)

dM (i, j-l, k) = M (i, j, k)-M (i, j-l, k)

dM (i-l, j + l, k) = M (i, j, k) -M (i-l, j + l, k)

dM (i + l, j-l, k) = M (i, j, k) -M (i + l, j-l, k)

dM (i + l, j, k) = M (i, j, k) _M (i + l, j, k)

dM (i + l, j + l, k) = M (i, j, k)-M (i + l, j + l, k)

dM (i-l, j-l, k + l) = (i, j, k) -M (i-l, j-l, k + l)

dM (i-l, j, k + l) = (i, j, k) ~ M (i-l, j, k + l)

dM (i-l, j + l, k + l) = M (i, j, k) -M (i-l, j + l, k + l)

dM (i, j-l, k + l) = (i, j, k) -M (i, j-l, k + l)

dM (i, j, k + l) = M (i, j, k) -M (i, j, k + l)

dM (i, j + l, k + l) = (i, j, k) -M (i, j + l, k + l)

dM (i + l, j-l, k + l) = M (i, j, k) -M (i + l, j-l, k + l)

dM (i + l, j, k + l) = M (i, j, k)-M (i + l, j, k + l)

dM (i + l, j + l, k + l) = M (i, j, k)-M (i + lJ + l, k + l), and when M (i, j, k)> 0 If all the difference values are positive, or if M (i, j, k) ≤ 0 and all the difference values are negative, the data point (i, j, k) is detected as a peak. [0079] Since the data point uses an integer, it may be far from the peak of the true peak depending on the data point density. In order to find a data point as close as possible to the true peak apex, for a data point detected as a peak, this data point and two data points adjacent to each other before and after each in the three directions of the X-axis, y-axis, and z-axis The maximum or minimum point of a quadratic function passing through a total of three points may be calculated, and the coordinates of the maximum or minimum point may be approximated as peak coordinates. In this case, the resulting data points are represented by real numbers, not integers.

[0080] The data points (i, j, k) detected as peaks are converted to the corresponding chemical shift values (C (n), C (n), C (n)) when described in the peak table. Is done.

H HC N

The peak table creation unit 10 assigns “n” (n is a natural number) as a peak ID to each detected peak. For each detected peak n, the peak table creating unit 10 calculates a peak ID, three-dimensional reference coordinates (C (n), C (n), C (n)), a half-value width (w (n), w (n

H HC N H HC

), w (n)) and peak intensity are made to correspond to each other to create a peak table. Peak stay

N

The table creation unit 10 transmits the created data of the peak table to the storage unit 12 and the corresponding mask peak search unit 13, respectively.

Next, creation of a mask file will be described. In order to create a mask file, the mask file creating section 11 reads out the data of the peak table stored in the storage section 12. Based on the peak table, the mask file creation unit 11 detects a group of peaks G (m) (m is a natural number) observed substantially linearly in the y-axis direction, and obtains a group of peaks G (m) Determine the X-z plane reference coordinates (P (m), P (m)) of. And each determined X-z plane

H N

A mask file that records the reference coordinates (P (m), P (m)) as the coordinates of the mask peak m

H N

create.

FIG. 10 shows a flowchart of creating a mask file according to the second embodiment. The mask file creating section 11 divides the Xz plane into sections having the same area, and determines whether or not each section contains a predetermined number or more of the reference coordinates of the peak (step S150). FIG. 11-1 is a projection diagram in which the reference coordinates of the peak recorded in the peak table are projected on the Xz plane. As shown in Fig. 11-1, the mask file creation unit 11 divides the Xz plane at a pitch of 0. Ippm in the X-axis direction and at a pitch of Ippm in the z-axis direction. Is searched for how many peak reference coordinates are included in each section.

[0084] The mask file creating unit 11 detects, as a peak group existence area, a section including a predetermined number or more of the reference coordinates of the peak (step S151). In Fig. 11-2, the hatched section is the section detected as the peak group existence area.

[0085] The mask file creation unit 11 finely adjusts the center coordinates (p (m), p (m)) of the section detected as the peak group existence area, and the number of corresponding peaks in the peak table is the largest. Tona

H N

Are determined as the reference coordinates (P (m), P (m)) of the peak group G (m) (step

H N

S152).

[0086] Whether or not the peak n corresponds to the center coordinates (p (m), p (m)) of the peak group existence area is determined.

H N

Can be determined using the function E (n, m) described above. That is, this function E (n

Η, Ν Η, Ν

, m), the peak η at which the evaluation value becomes 1 corresponds to the center coordinates (p (m), p (m))

H N

And the peak n at which the evaluation value is 0 means that the peak n corresponds to the mask peak m.

0 if | C _H (n)-p _H (m)-k _H I> t H + w _H (n)

or | C _N (n)-p _N (m)-k _N I> t _N + w _N (n)

1 if | C _H (n)-p _H (m)-k _H I <t _H + w _H (n)

[and | C _N (n) -p _N (m) -k _N <t _N + w _N (n)

Here, k and k are fine adjustment values of the peak n, t and t are thresholds set by the user, and w and w are

H N H N H N

This is the half-value width of n, which is the same as that described for the function E (n, m) described above.

Η, Ν

[0089] For optimization, fine adjustment values k, k were changed little by little, and evaluation was performed using all peaks.

H N

The total value of the fruit evaluation values is maximized. That is, k = Sk i, k = Sk j

H H N N

(Sk 0 · 002, Sk 0 · 02), and the values of i and j are 1 50, 1 49, 1 48

H N

By changing to ..47, 48, 49, 50, the x-z plane coordinates (C (n), C (n)) of peak n

H N

Search for the fine adjustment value for optimization within the range of ± 0.1 ppm in the X-axis direction and ± 1.0 ppm in the z-axis direction.

[0090] In the optimization, the total value S (i, j) of the evaluation values of the results of evaluation based on all peaks is calculated, Find (I, J) where this value is max S (i, j).

[0091] garden

S (ij) = ∑E¾ _N (n, m)

nN ₀

[0092] At this time, the fine adjustment values k = SkI and k = SkJ make the center coordinates (p (m), p (m)) the maximum.

H H N N H N

This is an optimization adjustment value for optimization. The reference coordinates (P (m), P (m)) of the peak group G (m) given by this optimization adjustment value are (p (m) + δ k I, ρ (m) + δ k J)

H N H H N N

[0093] The mask file creation unit 11 masks the reference coordinates (P (m), P (m)) of the peak group G (m).

H N

The coordinates of the peak m are stored in the mask file (step S153).

[0094] The noise filter device and the noise filter method according to the second embodiment described above can automatically create a peak table and a mask file, thereby further reducing the user's time burden in noise removal work. Can be. In addition, since oversight of the peak n and oversight of the peak group G (m) can be prevented, more accurate noise removal can be performed.

[0095] (Third embodiment)

Next, a third embodiment of the present invention will be described. The third embodiment is characterized in that a means for detecting and removing noise that cannot be completely removed in the first embodiment and the second embodiment is added. In the first and second embodiments, noise in which the mask peak m and the X-z plane coordinate difference are within a predetermined range is not discriminated as noise and remains in the peak table as it is. Was. The third embodiment is characterized in that such noise is detected and removed.

FIG. 12 is a block diagram illustrating the configuration of a noise filter device according to the third embodiment. Here, components having the same functions as those of the second embodiment are denoted by the same reference numerals, and description thereof will be omitted. The noise filter device according to the third embodiment is different from the noise filter device according to the second embodiment in that the noise filter device has a configuration similar to the noise filter device according to the second embodiment. Section 16, continuous peak group detection section 17, third noise removal section 18, A non-diagonal peak group detector 19, a code determiner 20, a fourth noise remover 21, a water noise detector 22, and a fifth noise remover 23 are provided.

[0097] The peak pair detection unit 15 detects a peak having a strong peak intensity symmetrically with respect to any one of three axes of the X-axis, the y-axis, and the z-axis around the peak. This is to detect one pair. The peak pair detection unit 15 transmits the peak IDs of the peaks forming the peak pair to the second noise removal unit 16 as noise data.

[0098] The peak pair detection section 15 mainly detects noise generated near a peak derived from a strong and sharp signal such as a methyl signal. The cause of noise near strong peaks, such as methyl signals, is that the heteronuclear decoupling performed during the time evolution of the proton nuclei during the spectral measurement process is slightly inadequate, and therefore the proton nuclei It is conceivable that peaks split by S-spin coupling and heterogeneous nuclear forces are mixed. Such noise is called incomplete decoupling noise. Further, in the measurement pulse scheme, a frequency-dependent delay time exists on each time axis. A sharp signal such as a methyl signal hardly relaxes during the delay time. Therefore, such a signal is recognized as a rectangular wave at the time of Fourier transform, and a sine waveform swells on the spectrum after the Fourier transform. This noise is multiplied by the squared cosine function or can be eliminated to some extent by the linear prediction method. It is difficult to completely eliminate the noise. Such noise is called wiggle noise. Fig. 13 shows an example of incomplete decoupling noise, and Fig. 14 shows an example of Uighur noise. As shown in FIG. 13 and FIG. 14, these noises tend to be found two by two symmetrically around the signal in the vicinity of the strong signal. The peak pair detection unit 15 detects such a pair of peaks, thereby efficiently detecting incomplete decoupling noise and wiggle noise.

[0099] The second noise removing unit 16 receives the noise data from the peak pair detecting unit 15. The second noise removing unit 16 reads the data of the peak table from the storage unit 12, removes the information of the peak corresponding to the noise from the peak table, and removes the information of the peak corresponding to the noise from the new peak. The table is transmitted to the storage unit 12.

[0100] The continuous peak group detection unit 17 places the y-axis near the diagonal peak around the diagonal peak. A continuous peak group consisting of peaks of the same sign that are observed while being continuously crowded in the direction is detected. The continuous peak group detection unit 17 transmits the peak IDs of peaks other than diagonal peaks included in the detected continuous peak group to the third noise removal unit 16 as noise data.

[0101] The continuous peak group detection unit 17 mainly detects noise that is observed as a methyl signal that is slow to relax and has a high intensity or a base line distortion generated near a side chain signal. The phase in the multidimensional NMR spectrum after Fourier transform is generally determined by the hardware or software delay time of the frequency-dependent expansion time set for each axis in the pulse scheme and the delay time generated when the NMR probe and receiver receive. Always a little off. These phase shifts are commonly used and can be fine-tuned by the user on spectral data processing software. In the presence of strong and slow-relaxing, methyl or side-chain signals, it is often very difficult to perfectly align these signals. A phase shift that cannot be completely adjusted distorts the baseline and causes many noise peaks. These types of noise are called tile noise. Fig. 15 shows an example of till noise.

[0102] The third noise removing unit 18 receives the noise data from the continuous peak group detecting unit 17. The third noise removing unit 18 reads the data of the peak table from the storage unit 12, removes the information of the peak corresponding to the noise from the peak table, and stores the new peak table from which the information of the peak corresponding to the noise is removed. Send to Part 12.

[0103] The off-diagonal peak group detection unit 19 detects a off-diagonal peak group other than a diagonal peak in which a predetermined number or more peaks are arranged substantially linearly in the y-axis direction from the diagonal peak, The information on the peaks included in the off-diagonal peak group is transmitted to the code determination unit 20.

[0104] The sign determination unit 20 determines which of the detected non-diagonal peak groups contains more positive or negative peaks, and determines the sign of the peak that is more contained as the sign of the off-diagonal peak group. . The sign determination unit 20 transmits the peak ID of a peak having a sign different from the sign determined for the off-diagonal peak group to the fourth noise removal unit 21 as noise data.

[0105] The off-diagonal peak group, which is also observed as a signal train with the diagonal peak force directed in the y-axis direction, is a signal of the same phase in principle, and the true NOE peak included in the off-diagonal peak group Is the same sign of the peak intensity. Therefore, peaks having different signs of the peak intensities are apparent noise. The sign determination unit 20 detects a peak having a different sign of the peak intensity as a different sign noise. Figure 16 shows an example of different sign noise

The fourth noise removing unit 21 receives the noise data from the code determining unit 20. The fourth noise removing unit 21 reads the data of the peak table from the storage unit 12, removes the information of the peak corresponding to the noise from the peak table, and creates a new peak table from which the information of the peak corresponding to the noise is removed. It is transmitted to the storage unit 12.

[0107] Water noise detection section 22 detects a peak corresponding to noise derived from light water from the power of the peak table. The noise originating from light water mainly appears in the light water signal, and usually appears as about 10,000 to 20000 noises in the range of ± 0.1-0.2 ppm. The water noise detection unit 22 detects a peak corresponding to this, and transmits the peak ID of the peak corresponding to the noise to the fifth noise removal unit 23 as noise data.

[0108] The fifth noise removing unit 23 reads the data of the peak table from the storage unit 12, removes the information of the peak corresponding to the noise from the peak table, and removes the information of the peak corresponding to the noise. The peak table is transmitted to the storage unit 12.

FIG. 17 shows a flowchart for removing noise by a noise filter device according to the third embodiment. Note that the same processes as those in the first embodiment are denoted by the same reference numerals, and description thereof will be omitted.

[0110] (Incomplete decoupling noise, Uighur noise removal)

After removing the noise by the mask filter method (step S16), the peak pair detection unit 15 reads the peak table from the storage unit 12, and detects the peak pair from the peak table (step S20).

FIG. 18 shows a flowchart of peak pair detection. First, the peak pair detecting unit 15 extracts a predetermined number of peaks as upper peaks in descending order of peak table force and peak intensity, and creates an upper peak table (step S200). The peak pair detection unit 15 assigns peak IDs “s” (s is a natural number) to the extracted upper peaks in descending order of peak intensity. Then, for each upper peak s, three-dimensional reference coordinates (C (s), C (s), C (s)) and half of each axis direction The upper peak peak value corresponding to the value range (w (s), w (s), w (s)) and the peak intensity I (s).

H HC N

Create a Bull. Fig. 19 shows an example of the upper peak table. The number of upper peaks to be extracted is a force that can be determined by the user. It is preferable to extract about 20 to 30 upper peaks in a normal three-dimensional NMR spectrum of a protein.

[0112] Next, for all the upper peaks included in the upper peak table, the peak force of the peak table also detects a peak pair. First, the upper peak for detecting a peak pair is set as the upper peak 1 having the highest peak intensity (step S201). Then, for the set upper peak, a peak pair is also searched for the intermediate strength of the peak table (step S203).

[0113] Peak pair detection is performed for two peaks that are located at coordinates that are symmetric with respect to any of the X-axis, y-axis, and z-axis directions and that have the same peak intensity. This is done by searching the peak table for a combination of peaks n and n.

1 2

[0114] Evaluation of whether two peaks n and n in the peak table are a peak pair

1 2

Is performed according to the following procedure. First, the peaks n and n are X

1 2

Determine whether the peak appears at a position that is symmetrical with respect to any of the x, y, and z axes. The evaluation of whether force is present on the y-axis line where peak n is almost the same as the upper peak s is given by the following function L (n, s), which results in 0 or 1. This function L (n, s)

The peak η at which the evaluation value is 1 due to Ν, Ν Η, ほぼ means that it is on the same y-axis line as the upper peak s, and the peak n at which the evaluation value is 0 is different from the upper peak s y It means that it exists on the axis line.

[0115] [Number 7]

0 if | C _H (n)-C _H (s) |> t _H + w _H (n) + w _H (s)

or | C _N (n)-C _N (s) |> t _N + w _N (n) + w _N (s)

1 if | C _H (n)-C _H (s) | <t _H + w _H (n) + w _H (s)

[and | C _N (n)-C _N (s) | <t _N + w _N (n) + w _N (s)

[0116] Similarly, for the x-axis and z-axis lines, the following functions L (n, s) and L

HC, N

Consider using (n, s).

[0117] [Number 8] 0 if | C _HC (n)-C _HC (s) |> t _HC + w _HC (n) + w _HC (s)

or | C _N (n)-C _N (s) |> t _N + w _N (n) + w _N (s)

1 if | C _HC (n) -C _HC (s) | <t HC ^{+ W} HC (n) + ^W HC ( ^S ) and | C _N (n)-C _N (s) | <t _N + w _N (n) + w _N (s)

[0118] [Number 9]

0 if | C _H (n)-C _H (s) |> t _H + w _H (n) + w _H (s)

o ^r | C _H c ( ⁿ )-C _HC (s) |> t _HC + w _HC (n) + w _HC (s)

1 if | C _H (n)-C _H (s) | <t H + w _H (n) + w _H (s)

and | C _HC (n) — C _HC (s) | ≤t HC + ^W HC (n) + w _HC (s)

[0119] The threshold values t 1, t 2, and t 3 used here are appropriately set by the user according to measurement conditions and the like.

H HC N

Can do. For example, to perform an appropriate noise search on spectra measured by NMR at 600-800 MHz, it is preferable to use the values 0.01 ppm, 0.02 ppm, and 0.1 ppm, respectively.

[0120] Then, using the function P (n, s), the peak n confirmed to exist on the y-axis line

HC

Evaluate that is the peak that is sufficiently close and far from the upper peak s on the y-axis line. The peak n whose evaluation value is 1 by this function P (n, s) is sufficient as the upper peak s on the y-axis line.

HC

It means that the peak is close to the distance, and the peak n whose evaluation value is 0 means that the peak exists at a position distant from the upper peak s.

[0121] [Number 10]

0 if | C _HC (n) — C _HC (s) |> 1 _HC + w _HC (n) + w _HC (s)

1 ^if | C He ( ⁿ ) — ^C HC ( ^s ) | ^≤ IHC + ^W HC ( ^N ) + ^W HC ( ^S )

[0122] The functions P (n, s) and P (n, s) are used in the same way for the x-axis and z-axis lines.

[0123] [Number 11] 0 if | C _H (n)-C _H (s) |> 1 _H + w _H (n) + w _H (s)

P _H (n, s) =

1 if | C _H (n)-C _H (s) | <1 _H + w H (n) + w _H (s)

[0124] [Number 12]

[0125] The thresholds 1, 1, and 1 used here are appropriately set by the user according to measurement conditions and the like.

H HC N

Can do. For example, to perform an appropriate noise search on spectra measured by NMR from 600 to 800 MHz, use the values of 1.Oppm, 1.Oppm, and 1.Oppm, respectively.

[0126] Next, two peaks n and n are arbitrarily selected from the peaks on the substantially same line in any of the X-axis, y-axis, and z-axis directions confirmed for the upper peak s. Choose

1 2 and two arbitrarily selected peaks n and n are based on the upper peak s on the y-axis line.

1 2

Consider positional symmetry. The two randomly selected peaks n and n are peaks on the y-axis line.

1 2

The function that evaluates the positional symmetry with respect to s is given by the function S (n, n, s). this

HC 1 2

The combination of peaks n and n for which the evaluation value is 1 by the function S (n, n, s) is the y-axis

HC 1 2 1 2

It means that the line has positional symmetry based on the upper peak s on the line, and a combination of peaks n and n with an evaluation value of 0 means that there is no positional symmetry.

1 2

[0127] [Number 13]

^S HC ( ⁿ l, ⁿ 2 'S):

[0128] The values of the evaluation thresholds rl and r2 here are values around 1 (rl <l, r2> l), and the forces that can be appropriately set by the user are 0.9 and 1.1, respectively. Is preferred.

[0129] Similarly, for the X-axis and z-axis lines, the functions S (n, n, s) and S (n, n, s)

Use H 1 2 N 1 2. [0130] [Number 14]

[0131] [Number 15]

S _N ( ⁿ l, ⁿ 2 ' ^S )

[0132] Further, it is evaluated whether two arbitrarily selected peaks n and n have intensity symmetry.

1 2

The two peaks n and n are evaluated by the function W (n, n) to determine whether they have intensity symmetry.

Given by 1 2 1 2 This function W (n, n) calculates the peaks n and n for which the evaluation value is 1.

1 2 1 2 combination means having intensity symmetry, peaks n and n for which the evaluation value is 0

The combination of 1 means that there is no intensity symmetry! /.

2

[0133] [Number 16]

W (n or n,) =

1 if ^I > _{vl and} < _v2

The values of the evaluation thresholds vl and v2 here are values around 1 (vl <1, ν2> 1), and the forces that the user can appropriately set are 0.8 and 1.2, respectively. Is preferred.

[0135] The peak pair detection unit 15 calculates a value obtained by multiplying all the functions in the X-axis, y-axis, and z-axis directions by the following equation, and calculates the X-axis, y-axis, and z-axis values. A combination of peaks n and n for which the multiplied value does not become 0 in any axis direction is detected as a peak pair.

[0136] [Number 17] y-axis: L _HN (n _l , s) * L _HN (n ₂ s) * P _HC (n ₁ s) * P _HC (n ₂ s) * S _HC (n ₁ , n ₂ s) * W (n ₁ , n ₂ ) ≠ 0 x-axis: L _HC , _N (n i s) * L _HC , _N (n ₂ s) * P _H (n i s) * P _H (n ₂ , s ) * S _H (n n ₂ , s) * W (n n ₂ ) ≠ 0 z-axis: L _HHC (n _1; s) * L _HHC (n _2! S) * P _N (n _1; s ) * P _N (n ₂ , s) * S _N (n ₁ , n ₂ , s) * W (n ₁ , n ₂ ) ≠ 0

[0137] When a peak pair is detected (Step S204 Yes), the peak pair detection unit 15 uses the peak IDs of the peaks n and n constituting the detected peak pair as noise data to perform second noise removal.

1 2

The information is transmitted to the unit 16 (step S205).

[0138] For the set upper peak s, when a peak pair is not detected (step S204 No) or when a peak pair is detected and noise data is transmitted (step S204 Yes, step S205), the peak pair is detected. The detecting unit 15 determines whether a peak pair has been searched for all upper peaks in the upper peak table (step S206). If a peak pair is still found in the upper peak table, and there is a higher peak (step S206 No), the upper peak table in which the strength of the upper peak table corresponds to the lower rank of the peak intensity by one level. Then, a peak is selected and set as an upper peak for detecting a peak pair (step S207). Then, the process of detecting a peak pair is repeated for the set upper peak (step S203). If all upper peaks in the upper peak table have been searched (step S206

Yes), the peak pair detection unit 15 determines that the peak pair detection is completed, and ends the peak pair detection processing.

[0139] The second noise removing unit 16 reads the data of the peak table from the storage unit 12, and removes the information of the peak corresponding to the noise from the peak table (FIG. 17, step S21). The second noise removing unit 16 transmits to the storage unit 12 a new peak table from which information on the peak corresponding to the noise has been removed.

[0140] (Til noise removal)

Next, the continuous peak group detection unit 17 reads the data of the peak table stored in the storage unit 12, detects a peak corresponding to a diagonal peak from the peak table, and detects the peak corresponding to the diagonal peak in the y-axis direction. Then, a continuous peak group in which a predetermined number or more of peaks having the same sign are arranged substantially on a straight line within a predetermined range is detected (step S22).

FIG. 20 shows a flowchart for detecting a continuous peak group. The continuous peak group detection unit 17 extracts a peak corresponding to the diagonal peak from the peak table, and File (step S220).

[0142] The evaluation of whether or not the force is such that the peak n is a diagonal peak is given by a function X (n). This

Η, Ν

The peak η at which the evaluation value becomes 1 by the function X (η) of means a diagonal peak,

Η, Ν

The peak η at which the evaluation value becomes 0 by this function X (η) is a non-diagonal peak

Η, Ν

I do.

[0143] [Number 18]

0 if | C _H (n)-C _HC (n) |> t + w _H (n) + w _HC (n)

_{Η Η} .Ν ( ^η ) = 1 if | C _H (n)-C _HC (n) | <t + w _H (n) + w _HC (n)

The threshold value t used here is preferably a value of 0.1 Olppm.

[0145] The continuous peak group detection unit 17 assigns "a" (a is a natural number) as a diagonal peak ID to a peak determined to be a diagonal peak, and assigns a diagonal peak ID, 3D reference coordinates (C (a), C (a), C (a)), FWHM in each axis direction (w (a), w (a), w (a)),

H HC N H HC N

A peak table is created in correspondence with the peak intensity I (a). Figure 21 shows an example of the diagonal peak taper.

[0146] Next, for each diagonal peak included in the diagonal peak table, a continuous peak group is detected in the order of diagonal peak IDs. First, a diagonal peak for detecting a continuous peak group is set to diagonal peak 1 (step S221). From the peak table, for the set diagonal peak, peaks having a predetermined number or more of the same sign within a predetermined range and directed in the y-axis direction from the diagonal peak continuously substantially linearly. The continuous peak group arranged is detected (step S222).

[0147] Whether the peak n is included in the continuous peak group with respect to the diagonal peak a is evaluated by the following method.

[0148] First, it is evaluated whether or not the peak n is arranged substantially linearly from the diagonal peak a in the y-axis direction. The evaluation of whether or not the force of the peak n is substantially linear on the diagonal peak a from the diagonal peak a in the y-axis direction is given by the following function L (n, a) that results in 0 or 1. This function

Η, Ν

ピーク The peak n whose evaluation value is 1 due to L (n, a) is directed from the diagonal peak a to the y-axis direction and

Η, Ν

Means that they are almost aligned on a straight line, and the peak η where the evaluation value is 0 is the diagonal peak a It does not exist on the same y-axis line.

[0149] [Number 19]

[0150] Similarly, for the x-axis and z-axis lines, the functions L (n, a) and L (n, a) are used.

HC, NH, HC

The

[0151] [number 20]

"0 if | C _HC (n)-C _HC (a) |> t _HC + _HC (n) + w _HC (a) or | C _N (n)-C _N (a) |> t _N + w _N (n) + w _N (a)

1 if | C _HC (n)-C _HC (a) | <t _HC + w _HC (n) + w _HC (a)

and | C _N (n)-C _N (a) | <t _N + w _N (n) + w _N (a)

[0152] [Number 21]

0 if | C _H (n)-C _H (a) |> t _H + w _H (n) + w _H (a)

^0r | ^C HC ( ⁿ )-C HC (a) |> ^l HC + w _HC (n) + w _HC (a)

1 if | C _H (n)-C _H (a) | <t _H + w _H (n) + w _H (a)

[and | C _HC (n)-C _HC (a) | <t _HC + w _HC (n) + w _HC (a) have thresholds t, t, and t of 0.01 ppm, 0.02 ppm, and 0, respectively. It is preferred to use lppm values.

Furthermore, it is evaluated whether the peak n exists within a predetermined range in the y-axis direction with respect to the diagonal peak a. The evaluation of whether or not the force of the peak n is within the predetermined range in the y-axis direction at the diagonal peak a is given by the following function D (n, a), which results in 0 or 1. Evaluated by the function D (n, a)

HC HC

Peak n with a value of 1 means that it is sufficiently close to diagonal peak a. Peak n with an evaluation value of 0 means that the distance from diagonal peak a is long, and The possibility that the noise is a tilt noise of a is denied.

[Number 22]

^{0 if} | ^C Hc ( ⁿ ) ^-C Hc ( ^a ) |> d _HC + w _HC (n) + w _HC (a)

D _HC (n, a) =

1 if | C _HC (n) — C _HC (a) | ≤d _HC + ^W HC (n) + w _HC (a)

[0156] Similarly, functions D (n, a) and D (n, a) are used for the x-axis and z-axis lines.

H N

[0157] [Number 23]

0 if | C _H (n)-C _H (a) |> d _H + w _H (n) + w _H (a)

D _H (n, a) =

{1 if | C _H (n)-C _H (a) | <d _H + w _H (n) + w _H (a)

[0158] [Number 24]

0 if | C _N (n)-C _N (a) |> d _N + w _N (n) + w _N (a)

D _N (n, a) =

1 if | C _N (n)-C _N (a) | ≤d _N + w _N (n) + w _N (a)

Here, d, d, and d are thresholds, which can be set by the user. For example, 600-80

H HC N

To perform an appropriate noise search on the spectrum measured by OMHz NMR, 3. Oppm, 3. Oppm, 3. Oppm is preferred!

[0160] Here, for the diagonal peak a, the peak group where L (n, a) * D (n, a) = 1 is peaked.

H, N HC

Set peak group r, assign peak IDs rl, r2, ... for the peaks contained in peak group r, and evaluate whether the peaks contained in peak group r overlap continuously.

[0161] The evaluation of whether the peak rl and the peak r2 included in the peak group r overlap on the y-axis is given by the following function O (rl, r2) that results in 0 or 1.

[0162] [Number 25]

0 if | C _HC (rl)-C _HC (r2) |> v _HC + w _HC (rl) + w _HC (r2)

0 _HC (rl, _r 2) =

1 if | C _HC (rl)-C _HC (r2) | <v _HC + w _HC (rl) + w _HC (r2)

[0163] To similarly consider the x-axis and z-axis lines, L (n, a) * D (n, a) = 1 or L (n, a) * D (n, a) = 1

H, HC N

For the peak rl and peak r2 included in the peak group r, the following functions O (rl, r2) and 0 (rl, r

H N

The continuous peak group is detected in the same manner as in the case of the y-axis using 2).

[Number 26]

0 if | C _H (rl)-C _H (r2) |> v _H + w _H (rl) + w

0 _H (rl, r2) =

1 if | C _H (rl)-C _H (r2) | <v _H + w _H (rl) + w

[0165] [Number 27]

0 if | C _N (rl) -C _N (r2) |> v _N + w _N (rl) + w _N (r2)

0 _N (rl, r2) =

.1 if | C _N (rl)-C _N (r2) | <v _N + w _N (rl) + w _N (r2)

[0166] Here, the thresholds v, V, and V for the degree of overlap between peaks are values set by the user,

H HC N

For example, it is preferable to use 0.01 ppm, 0.03 ppm, and 0.1 ppm in a 600 MHz NMR apparatus. A third peak r3 overlapping with the peak rl or the peak r2 is similarly searched for the searched peak set rl. By repeating this operation, peaks rl, r2, r3, and r4 are searched for the peaks r that overlap continuously. If the total number of consecutive overlapping peaks exceeds the reference value (usually set to 8), it is detected as a continuous peak group.

[0167] When a continuous peak group is detected (Step S223 Yes), the continuous peak group detection unit 17 determines the peak IDs of all the peaks included in the detected continuous peak group except for the diagonal peaks as noise. The data is transmitted to the third noise removing unit 18 as data (step S224).

[0168] For the set diagonal peak a, if a continuous peak group is not detected (step S223 No) or if a continuous peak group is detected and noise data is transmitted (step S223 Yes, step S224), the continuous The peak group detection unit 17 determines whether a continuous peak group has been searched for all diagonal peaks in the diagonal peak table (step S225). If the continuous peak group is still searched in the diagonal peak table, and there is a diagonal peak (step S225 No), the diagonal peak ID is increased by one and the diagonal to search for the continuous peak group is increased. Set as peak (step S226). Then, a continuous peak group is searched for the set diagonal peak (step S222). All pairs in the diagonal peak table If a continuous peak group has been searched for the angular peak (Step S225 Yes), the continuous peak group detection unit 17 determines that the continuous peak group search has been completed and ends the continuous peak group search processing.

[0169] Here, the force obtained by searching a continuous peak group for all the diagonal peaks in the diagonal peak table, among the diagonal peaks included in the diagonal peak table, the strongest peak intensity V and the diagonal peak. You can also search for a group of continuous peaks only! Searching for consecutive peak groups The number of diagonal peaks can be set by the user. Usually, however, it is preferable to search the top 20 to 30 diagonal peaks in descending order of peak intensity! / ,.

[0170] The third noise removing unit 18 reads the data of the peak table from the storage unit 12, and removes the information of the peak corresponding to the noise from the peak table (FIG. 17, step S23). The third noise removing unit 18 transmits to the storage unit 12 a new peak table from which information on the peak corresponding to the noise has been removed.

[0171] (Removal of different sign noise)

Next, the non-diagonal peak group detection unit 19 reads the data of the peak table stored in the storage unit 12, extracts the peak corresponding to the diagonal peak also in the peak table, and calculates the peak from the diagonal peak in the y-axis direction. Then, a non-diagonal peak group in which a predetermined number or more of peaks are arranged substantially on a straight line is detected (FIG. 17, step S24).

FIG. 22 shows a flowchart of detecting a non-diagonal peak group. Note that the same processes as those in the flowchart of detecting the continuous peak group described in FIG. 20 are denoted by the same reference numerals, and description thereof will be omitted.

First, the off-diagonal peak group detection unit 17 sets the diagonal peak for detecting the off-diagonal peak group to diagonal peak 1 (step S221). Then, from the peak table, for the set diagonal peaks, a non-diagonal peak group in which a predetermined number or more of peaks are arranged substantially linearly in the y-axis direction from the diagonal peak is detected (step S250). ).

[0174] The following method is used to evaluate whether peak n is included in the off-diagonal peak group for diagonal peak a.

First, it is evaluated whether or not the peak n is arranged substantially linearly from the diagonal peak a in the y-axis direction. Peak n is aligned almost linearly from diagonal peak a in the y-axis direction. The evaluation of the presence or absence of the force is given by the following function L (n, a), which results in 0 or 1. This function

Η, Ν

Means that the peak η where the evaluation value is 0 does not exist on the same y-axis line as the diagonal peak a.

[0176] [Number 28]

"0 if | C _H (n)-C _H (a) |> t _H + w _H (n) + w _H (a)

or | C _N (n)-C _N (a) |> t _N + w _N (n) + w _N (a)

^L H, N (n, a) =

1 if | C _H (n)-C _H (a) | <t _H + w _H (n) + w _H (a)

and | C _N (n)-C _N (a) | <t _N + w _N (n) + w _N (a)

[0177] The threshold values t and t used here can be appropriately set by the user. For example, 600-8

H N

To perform an appropriate noise search on the spectrum measured by OOMHz NMR, it is preferable to use 0.01 ppm and 0.1 ppm, respectively.

[0178] The off-diagonal peak group detection unit 17 detects a group of peaks other than the diagonal peak a on the same y-axis line as the diagonal peak a as a off-diagonal peak group for the diagonal peak a, and For the X peaks included in the off-diagonal peak group, assign _r , r,

1 2

The data is transmitted to the code determination unit 20 (step S251).

[0179] The sign determination unit 20 determines which of the positive and negative peaks n the off-diagonal peak group excluding the diagonal peaks includes, and determines the sign of the most contained peak n as the sign of the off-diagonal peak group. Is determined (step S252).

[0180] The sign of the off-diagonal peak group is determined by the following determination equation S (a).

HC

[0181] [Number 29]

1 if L _HN (^ ^) = 1 811 (1 1 (Γ;)> 0

PHc (r,,) = S _HC (a) = ∑Ρ η, &)

1 1 if L _HN (Γ;, a) = 1 and Ι () <0 xi = i

[0182] The sign peaks positively when S (a)> 0.7, or negative sign when S (a)> 0.7

HC HC

The correct sign of the group is determined. | S (a) If I ≤0.7, sign cannot be determined

HC

become. At this time, when 70% or more of the individual peaks show the same sign, The sign of the peak is determined as the sign of the off-diagonal peak group.

[0183] If the sign of the off-diagonal peak group is positive (Step S253 Yes), the sign determination unit 20 determines the peak ID of the peak having the negative peak intensity from among the peaks included in the off-diagonal peak group. The data is transmitted to the fourth noise removing unit 21 as noise data (step S254).

[0184] If the sign of the off-diagonal peak group is negative (No in step S253), the sign determination unit 20 determines the peak ID of the peak having a positive peak intensity from among the peaks included in the off-diagonal peak group. The data is transmitted to the fourth noise removing unit 21 as noise data (step S255).

The off-diagonal peak group detection unit 19 determines whether the off-diagonal peak group has been searched for all the diagonal peaks in the diagonal peak table (step S256). Search the non-diagonal peak group in the diagonal peak table yet! / ヽ If there is a diagonal peak (No in step S256), increase the diagonal peak ID by one and delete the non-diagonal peak group. Set as the diagonal peak to search (step S226). Then, the non-diagonal peak group detection unit 19 detects a non-diagonal peak group for the set diagonal peak (step S250). If the non-diagonal peak group has been searched for all the diagonal peaks in the diagonal peak table (step S2 56 Yes), the non-diagonal peak group detection unit 19 determines that the non-diagonal peak group detection has been completed. Then, the off-diagonal peak group detection processing ends.

[0186] The fourth noise removing unit 21 reads the data of the peak table from the storage unit 12, and removes the information of the peak corresponding to the noise from the peak table (FIG. 17, step S26). The fourth noise removing unit 21 transmits a new peak table from which information of the peak corresponding to the noise has been removed to the storage unit 12.

[0187] (Water noise removal)

Next, the water noise detection unit 22 reads the data of the peak table stored in the storage unit 12, and detects a peak corresponding to the noise derived from light water in the peak table (FIG. 17, step S27).

[0188] The water noise detector 22 detects the light water signal observed near C = 4.7 ppm.

H

The center X coordinate is determined, and a peak whose peak table force X coordinate is within the range of this center X coordinate force predetermined threshold value is detected. Here, the threshold can be appropriately set by the user, but is preferably set in the range of 0.1 to 0.2 ppm. [0189] The water noise detection unit 22 determines the peak corresponding to the light water-derived noise as the noise in the peak table as noise, and sends the peak ID of the peak corresponding to the noise to the fifth noise removal unit 23 as noise data. Send.

[0190] The fifth noise removing unit 23 reads the data of the peak table from the storage unit 12, and removes the information of the peak corresponding to the noise from the peak table (FIG. 17, step S28). The fifth noise removing unit 23 transmits to the storage unit 12 a new peak table from which information on the peak corresponding to the noise has been removed.

[0191] FIG. 23- 1, FIG. 24 1, Fig. 25 1, Fig. 26-1 is a partially enlarged view of the ¹ H- ¹³ N HSQC- NOESY space Tuttle in FIG.

Figure 23-1 is an enlarged view of the part including incomplete decoupling noise and Uyghur noise, where X corresponding to the incomplete decoupling noise or Uighur noise and * for the NOE peak. Fig. 24-1 is an enlarged view of a part including the tile noise, and an X is added to the peak corresponding to the tile noise. Fig. 25-1 is a partially enlarged view including the sign noise, where the peak corresponding to the sign noise is marked with X, and the NOE peak is marked with *. FIG. 26A is a partially enlarged view including water noise, and an arrow is attached to the boundary line of the existence region of the water noise.

Figure 23-2, Figure 24-2, Figure 25-2, and Figure 26-2 show the third embodiment from Figure 23-1, Figure 24-1, Figure 25-1, and Figure 26-1, respectively. FIG. 4 is a diagram after noise has been removed by a noise filter device. Fig. 23-2, Fig. 24-2, Fig. 25-2, and Fig. 26-2 are compared with Fig. 23-1, Fig. 24-1, Fig. 25-1, and Fig. 26-1, respectively. It can be seen that the incomplete decoupling noise, the Uighur noise, the Til noise, the different sign noise, and the ゥ -outer noise are efficiently removed by the noise filter device.

The noise filter device and the noise filter method according to the third embodiment described above can be completely removed by a mask filter method, such as incomplete decoupling noise, wiggle noise, tile noise, different sign noise, and water noise. Noise can be removed efficiently and accurately before signal assignment.

In the third embodiment, (1) off signal noise, (2) incomplete decoupling noise, wiggle noise, (3) tile noise, (4) different sign noise, and (5) war signal noise The order of removing noise is not limited to this. It goes without saying that the order of removing noise can be changed as appropriate for efficient noise search.

[0194] Further, the noise filter method described in Embodiments 1 to 3 can be provided as a noise filter program for causing a computer to execute the noise filter method. The noise filter program according to the first to third embodiments is a computer-readable recording medium such as a CD-ROM, a floppy (R) disk (FD), or a DVD in an installable or executable file. It is provided by being recorded on a medium. Further, the noise filter program according to the first to third embodiments may be stored on a computer connected to a network such as the Internet, and provided by being downloaded via the network. ,.

Industrial applicability

[0195] As described above, the noise filter device, the noise filter method, and the noise filter program according to the present invention can be easily performed without requiring signal assignment in a three-dimensional heteronuclear correlation NMR ^ vector. In addition, noise can be removed in a short time. The noise filter device, the noise filter method, and the noise filter program of the present invention having powerful features are extremely useful in the field of protein NMR three-dimensional structure analysis.

Claims

The scope of the claims

[1] X-axis corresponding to chemical shift value C of ¹ H nucleus directly observed, obtained by time expansion ¹

H

Chemical shift values of heterogeneous nuclei other than H nuclei The z-axis corresponding to C and other parameters

X

A noise filter device for removing three-dimensional heteronuclear correlation NMR ^ vector color noise corresponding to the three-axis color of the y-axis,

In the three-dimensional heteronuclear correlation NMR spectrum, the peaks G (m) (m is a natural number) of the peak group G (m) (m is a natural number) in which a predetermined number or more peaks are observed in a substantially straight line in the axial direction of the y-axis. — Define the z-plane reference coordinates (P (m), P (m)) as the coordinates of the mask peak m.

H X

It has a mask file in which the coordinates of the peak m are recorded in advance,

For each peak n (n is a natural number) in the three-dimensional heteronuclear correlation NMR spectrum, a peak corresponding to the X-z plane reference coordinates (C (n), C (n)) of each peak n for each peak n

H X

Storage means for storing a table;

For each peak n in the peak table, the X-z plane reference coordinates (C (

H

n), C (n)) and the coordinates (P (m), P (m)) of each mask peak m in the mask file

X H X

Calculate the difference (C (n) -P (m), C (n) -P (m)), and enter the coordinate difference within a predetermined range.

H H X X

Corresponding mask peak searching means for searching for a corresponding mask peak as a corresponding mask peak of this peak n;

The corresponding mask peak searching means determines that a peak that cannot be searched for a corresponding mask peak is noise, and removes information on the peak determined to be noise from the peak table stored in the storage means. A first noise removing means,

A noise filter device comprising:

[2] The noise filter device according to claim 1, further comprising a peak table creating means for creating the peak table.

[3] The noise filter device according to claim 1 or 2, further comprising a mask file creating means for creating the mask file.

[4] The noise filter device according to any one of claims 13 to 13, wherein the predetermined range is set corresponding to a half width of each peak n.

[5] The z-axis is an axis corresponding to a chemical shift value of a ¹⁵ N nucleus or a ¹³ C nucleus. The noise filter device according to claim 14.

[6] The three-dimensional heteronuclear correlation NMR spectrum is a three-dimensional HSQC-NOESY spectrum, and the y-axis corresponds to a chemical shift value C of ¹ H nucleus obtained by time expansion.

HC

The noise filter device according to any one of claims 11 to 15, wherein:

[7] The peak table force is also the intensity of the peak intensity, and a predetermined number of peaks are sequentially extracted as upper peaks, and for each of the extracted upper peaks, the X-axis, y-axis, and z-axis are centered on the upper peak. Peak pair detecting means for detecting a peak pair of a pair of peaks present at coordinates substantially symmetric with respect to any one of the three axis directions and having substantially the same peak intensity,

A second noise removing unit that determines peaks forming the peak pair as noise, and removes information of the peak determined as noise from the peak table stored in the storage unit;

17. The noise filter device according to claim 16, comprising:

[8] Peaks having substantially the same values of the X and y coordinates are extracted from the peak table as diagonal peaks, and for each of the extracted diagonal peaks, the X axis, y axis, and A continuous peak group detecting means for detecting, as a continuous peak group, a group of peaks having consecutive peaks having the same sign in any one of the three axis directions of the z axis and the z axis;

Third noise removing means for determining peaks other than diagonal peaks included in the continuous peak group as noise, and removing information of the peak determined as noise from the peak table stored in the storage means;

The noise filter device according to any one of claims 17 to 17, further comprising:

[9] From the peak table, peaks having substantially the same values of the X coordinate and the y coordinate are extracted as diagonal peaks, and for each extracted diagonal peak, a predetermined value is determined from the diagonal peak in the axial direction of the y axis. A non-diagonal peak group detecting means for detecting peak groups other than diagonal peaks in which peaks equal to or more than a number are arranged substantially linearly as non-diagonal peak groups;

For each detected off-diagonal peak group, the off-diagonal peak group Sign determining means for judging whether or not the peak is included a lot, and determining the sign of the peak that is included as a sign of the off-diagonal peak group;

The sign determined by the sign determining means for the off-diagonal peak group is compared with the sign of each peak included in the off-diagonal peak group, and is different from the sign determined for the off-diagonal peak group. Fourth noise removing means for removing a peak having a sign as noise from the peak table;

The noise filter device according to any one of claims 5 to 8, further comprising:

[10] The noise according to any one of [11] to [19], further comprising fifth noise removing means for removing a peak corresponding to noise derived from light water from the peak table. Filter device.

[11] X-axis corresponding to the chemical shift value C of ¹ H nucleus directly observed, obtained by time expansion ¹

H

X

A noise filter method for removing three-dimensional heteronuclear correlation NMR ^ vector color noise corresponding to the three y-axis colors,

From the actually measured three-dimensional heteronuclear correlation NMR spectrum, a peak having a predetermined peak intensity or more is extracted as a peak n (n is a natural number), and for each extracted peak, the X-z plane reference coordinates of that peak n ( C (n), C (n)) corresponding peak tape to create a peak table

H X

File creation step,

A storing step of storing the peak table;

In the three-dimensional heteronuclear correlation NMR spectrum, a peak group G (m) (m is a natural number) in which more than a predetermined number of peaks are observed in a substantially straight line in the axial direction of the y-axis. Then, the X-z plane reference coordinates (P (m), P (m)) of each peak group G (m) are determined and determined.

H X

Record the specified X-z plane reference coordinates (P (m), P (m)) as the coordinates of the mask peak m.

H X

Creating a mask file for creating a mask file,

For each peak n in the peak table, the X-z plane reference coordinates (C (

H

n), C (n)) and the coordinate difference (C (n) —P (m) between the coordinates (P (m), P (m)) of each mask peak m.

X H X H H

(m), C (n) —P (m)), and the coordinate difference is determined in advance from the force of the mask file. A corresponding mask peak search step of searching for a mask peak falling within the determined range as a corresponding mask peak of this peak n;

In the corresponding mask peak search step, a peak for which a corresponding mask peak could not be searched is determined as noise, and information on the peak determined as noise is removed from the peak table stored in the storage step. A noise filtering method, comprising: a first noise removing step.

X-axis corresponding to chemical shift value C of ¹ H nucleus directly observed, obtained by time expansion ¹

H

X

A program for causing a computer to execute a noise filter method for removing three-dimensional heteronuclear correlation NMR ^ vector color noise corresponding to the three-axis color of the y-axis,

H X

File creation step,

A storing step of storing the peak table;

H X

Creating a mask file for creating a mask file,

For each peak n in the peak table, the X-z plane reference coordinates (C (

H

X H X H H

(m), C (n) —P (m)), and the coordinate difference is determined in advance from the force of the mask file.

X X

A corresponding mask peak search step of searching for a mask peak falling within the determined range as a corresponding mask peak of this peak n;

In the corresponding mask peak search step, a peak for which a corresponding mask peak could not be searched is determined as noise, and information on the peak determined as noise is removed from the peak table stored in the storage step. A first denoising step; A noise filter program that causes a computer to execute the program.