CN107092790B - Electron cryo-microscopy three-dimensional density figure resolution detection method - Google Patents

Abstract

The invention provides a method for detecting the resolution of a three-dimensional density map of a cryo-electron microscope, comprising the steps of: inputting a three-dimensional density map of a cryo-electron microscope and a three-dimensional density map of pure noise; generating a mask set; Obtain the first intra-membrane data and the first extra-membrane data; calculate the first intra-membrane noise spectral power and the first extra-membrane noise spectral power; calculate and obtain a linear parameter set; use each mask to segment the cryo-EM 3D density map respectively. Obtain the second in-film data and the second out-of-film data; calculate and obtain the second in-film data spectrum signal-to-noise ratio curve corresponding to each mask according to the second in-film data, the second out-of-film data and the linear parameter set; The second in-film data spectrum signal-to-noise ratio curve graph obtains a resolution set, and draws a three-dimensional surface map of the resolution set according to the resolution set; obtains a global resolution value. The method for detecting the resolution of the three-dimensional density map of the cryo-electron microscope of the present invention has the advantages of simplicity, intuition and convenient operation.

Description

Cryo-EM 3D Density Map Resolution Detection Method

technical field

The invention relates to the field of image processing, in particular to a method for detecting the resolution of a three-dimensional density map of a cryo-electron microscope.

Background technique

As an emerging technology, cryo-electron microscopy (Cryo-EM) technology has developed rapidly in recent years and has achieved remarkable achievements. More and more macromolecules are reconstructed by cryo-electron microscopy technology to obtain sub-nanometer three-dimensional structures. Traditional three-dimensional reconstruction techniques in the field of structural biology, such as x-ray three-dimensional reconstruction technology and nuclear magnetic resonance three-dimensional reconstruction technology (NMR), have their own inherent defects, such as x-ray technology can only study crystals, not non-ferrous metals. Crystals undergo three-dimensional reconstruction; NMR techniques cannot perform three-dimensional reconstruction of molecules with large molecular weights (over 100KDa). In contrast, cryo-electron microscopy has significant advantages: (1) there is no crystallization requirement for the research sample, and three-dimensional reconstruction can be performed on both crystal and amorphous; (2) there is no molecular weight limit for the research sample, and it can be used for large molecular weight Three-dimensional reconstruction of biomolecules; (3) Rapid freezing of the sample, so that the water in the sample is in a glass state, and the biomolecules are very close to the physiological state; (4) The biomolecules in the sample are in different stages of the physiological process, which can be used to study physiological The dynamic evolution of biomolecules in the process, etc. The remarkable advantages of cryo-EM technology make it have broad development prospects.

The main method for 3D reconstruction using cryo-EM is single particle reconstruction (SPR). The process of single particle reconstruction is as follows: (1) Frozen sample preparation; (2) Electron microscope image acquisition; (3) Electron microscope image processing, including: CTF correction, particle selection, etc.; (4) 3D reconstruction, including particle classification , averaging, building an initial model, iteratively optimizing the final model, etc. After single particle reconstruction obtains the 3D density map of biomolecules, it is of great significance to evaluate the quality of the reconstructed model, that is, resolution detection: it helps to compare the advantages and disadvantages of reconstruction algorithms, and promotes the integration and communication of different technologies. Promote the development of cryo-electron microscopy technology, promote the development of structural biology, etc.

At present, the commonly used resolution detection methods include Fourier shell correlation (FSC) method and Spectral signal-to-noise ratio (SSNR) method. To detect the resolution by the FSC method, the particle dataset needs to be randomly divided into two half datasets, three-dimensional reconstruction of the two half datasets is performed, and then the correlation between the reconstruction results of the two half datasets in the Fourier space is calculated, The FSC curve is obtained, and the final global resolution is obtained by threshold truncation. Using the FSC method for resolution detection has the following disadvantages: (1) What this method detects is actually the repeatability of the reconstruction algorithm. Different from the traditional resolution detection, the selection of the threshold has no intuitive meaning. (2) This method is easily affected by noise. Since there is a lot of noise in the high-frequency part, this method often calculates the correlation of the noise as the particle correlation when calculating, and the phenomenon of over-fitting occurs; (3) This method needs to divide the particle data set into two half data sets and perform 3D reconstruction respectively. The reconstruction quality of the half data set is lower than that of the whole data set. The resolution of the structural results does not match, and the calculation process is redundant.

The SSNR method calculates the signal-to-noise power ratio of the three-dimensional density map in the frequency domain, and then truncates it through a threshold to calculate the global resolution. Compared with the FSC method, the threshold used in calculating the resolution has a clear meaning. As usual, 1 is taken as the threshold. At this time, the signal power is equal to the noise power. The existing method for calculating SSNR is to re-project the 3D density map. The image in the projected space is considered as the denoised image, subtract the projection image from the original image to obtain the noise image, and then use the noise image obtained from the pure noise 3D density map. The noise attenuation factor is used to correct the signal power and noise power of the projected image. Finally, the spectral signal-to-noise ratio of the three-dimensional density map is obtained from the corrected noise power of the projected image signal power. Using the existing method to calculate the SSNR is to consider the signal-to-noise ratio of the three-dimensional space in the two-dimensional space, which is not intuitive. The calculation result depends on the re-projection process, and the original data set is required in the calculation process.

The cryo-electron microscope technology is in the stage of rapid development and has broad application prospects. However, the existing cryo-electron microscopy three-dimensional density map resolution detection methods have major defects, so it is necessary to propose a new resolution detection method.

SUMMARY OF THE INVENTION

In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for detecting the resolution of a three-dimensional density map of a cryo-electron microscope, which directly calculates the spectral signal-to-noise ratio of a three-dimensional density map in a three-dimensional space, does not require a reprojection process, and has simple, intuitive, and The advantage of convenient operation.

In order to achieve the above purpose, the present invention provides a method for detecting the resolution of a three-dimensional density map of cryo-electron microscopy, comprising the steps of:

S1: input a three-dimensional density map of cryo-electron microscope and a three-dimensional density map of pure noise of the three-dimensional density map of cryo-electron microscope;

S2: generate a mask set according to the three-dimensional density map of the cryo-electron microscope, where the mask set includes n masks of different sizes, where n is an integer greater than zero;

S3: Segment the pure noise three-dimensional density map by using each of the masks to obtain a first in-film data and a first out-of-film data corresponding to each of the masks of the pure noise three-dimensional density map, respectively. ;

S4: Calculate a first in-film noise spectral power and a first out-of-film noise spectral power corresponding to each of the masks according to the first in-film data and the first out-of-film data;

S5: Calculate and obtain a linear parameter set λ _p according to the first in-film noise spectral power and the first out-of-film noise spectral power in each group;

where 0≤p≤n, and p is an integer, is the first in-film noise spectral power corresponding to the p-th mask, is the first out-of-film noise spectral power corresponding to the p-th mask; λ _p is the linear fitting parameter;

S6: Segment the cryo-EM 3D density map by using each of the masks to obtain a second intra-membrane data and a second extra-membrane data corresponding to each mask of the cryo-EM 3-D density map, respectively. ;

S7: Calculate and obtain a spectral signal-to-noise ratio curve of the second in-film data corresponding to each mask according to the second in-film data, the second out-of-film data and the linear parameter set;

S8: obtaining a resolution set according to the second in-film data spectrum signal-to-noise ratio curve, and drawing a three-dimensional surface graph of the resolution set according to the resolution set;

S9: Calculate and obtain a global resolution value of the three-dimensional density map of the cryo-electron microscope according to the three-dimensional surface map of the resolution set.

Preferably, the mask adopts a binary mask, the binary mask includes a mask area, the value inside the mask area is 1, and the value outside the mask area is 0.

Preferably, in the step S2, the three-dimensional cryo-electron microscope is processed by setting and adjusting a density isosurface threshold parameter of a three-dimensional density map and an outwardly expanding Gaussian layer parameter in a transmission electron microscope image processing software. density map to obtain the mask set.

Preferably, the S3 step further comprises the steps:

Multiplying the value of the pure noise three-dimensional density map on each coordinate in the three-dimensional space by the value on the corresponding coordinate of the current mask to obtain the first in-film data corresponding to the current mask;

subtracting the first in-film data from the pure noise three-dimensional density map to obtain the first out-of-film data corresponding to the current mask;

The steps are repeated until the first in-film data and the first out-of-film data corresponding to all the masks are obtained.

Preferably, the S4 step further comprises the steps:

Converting the first in-film data corresponding to the current mask to a first frequency space through Fourier transform; converting the first out-of-film data corresponding to the current mask through Fourier transform convert to a second frequency space;

A spherical shell model shell(s) is established in the first frequency space and the second frequency space respectively, where s is the radius of the spherical shell model;

Calculate the first in-film noise spectral power and the first out-of-film noise spectral power corresponding to the current mask according to a formula (2):

Wherein, RPS(s) is the spectral power; when RPS( _s ) is the spectral power of the noise in the first membrane, Ks is the coordinates of the point on the spherical shell model in the first frequency space, and the N _s is the number of points on the spherical shell model in the first frequency space, and M(K _s ) is the Fourier transform of the data in the first membrane;

When RPS(s) is the spectral power of the first out-of-membrane noise, K _s is the coordinates of a point on the spherical shell model in the second frequency space, and N _s is the inner The number of points on the spherical shell model, M(K _s ) is the Fourier transform of the first out-of-membrane data.

Preferably, the S7 step further comprises the steps of:

S71: Convert the second in-film data corresponding to the current mask to a third frequency space through Fourier transform; convert the second out-of-film data corresponding to the current mask through Fourier transform The leaf transform transforms into a fourth frequency space;

S72: respectively establish a spherical shell model shell(s) in the third frequency space and the fourth frequency space, where s is the radius of the spherical shell model;

S73: Calculate the spectral power of the sum of a second in-film signal and noise and a second out-of-film noise spectral power corresponding to the current mask according to the formula (2):

When RPS(s) is the spectral power of the sum of signal and noise in the second membrane, K _s is the coordinates of a point on the spherical shell model in the third frequency space, and N _s is the The number of points on the spherical shell model in the third frequency space, M(K _s ) is the Fourier transform of the data in the second membrane;

When RPS(s) is the spectral power of the second out-of-membrane noise, K _s is the coordinates of a point on the spherical shell model in the fourth frequency space, and N _s is the inner The number of points on the spherical shell model, M(K _s ) is the Fourier transform of the second out-of-membrane data;

S74: Using a formula (3), the linear parameter set λ _p and the second out-of-film noise spectral power to estimate a second in-film noise spectral power of the cryo-EM three-dimensional density map:

RPS _np (s) = λ _p · RPS _op (s) (3);

Wherein, RPS _np (s) is the second in-film noise spectral power corresponding to the pth mask; RPS _op (s) is the second out-of-film noise spectral power corresponding to the pth mask;

S75: Calculate the spectral power of the second in-film signal by subtracting the spectral power of the second in-film noise from the summed spectral power of the second in-film signal and noise to obtain a second in-film signal spectral power RPS _sp (s ):

RPS _sp (s) = RPS _ip (s) - RPS _np (s) (4);

Wherein, RPS _ip (s) is the spectral power of the sum of the second in-film signal and noise corresponding to the p-th mask; RPS _np (s) is the second in-film noise corresponding to the p-th mask spectral power;

S76: Calculate a signal-to-noise ratio of a second in-film data spectrum according to the second in-film signal spectral power and the second in-film noise spectral power, and draw the second in-film data spectral signal corresponding to the current mask Noise ratio curve;

S77: Repeat steps S71-S76 until the second in-film data spectrum signal-to-noise ratio curves corresponding to all the masks are obtained.

Preferably, the signal-to-noise ratio of the data spectrum in the second film satisfies the formula (5):

Wherein, SSNR _p (s) is the signal-to-noise ratio of the second in-film data spectrum corresponding to the p-th mask.

Preferably, in the step of drawing the second in-film data spectrum signal-to-noise ratio curve graph corresponding to the current mask: taking the logarithm of the second in-film data spectrum signal-to-noise ratio to obtain the second film The ordinate of the signal-to-noise ratio curve of the data spectrum in the second film, the radius of the spherical shell model corresponding to the signal-to-noise ratio of the data spectrum in the second film is taken as the horizontal axis of the signal-to-noise ratio curve of the data spectrum in the second film coordinates to form a signal-to-noise ratio curve diagram of the data spectrum in the second film.

Preferably, the S8 step further includes the steps:

set a threshold;

At least one coordinate point to be selected whose ordinate is equal to the threshold value on the signal-to-noise ratio curve of each second in-film data spectrum is intercepted in each of the second in-film data spectrum signal-to-noise ratio curves, and Selecting the candidate coordinate point with the smallest abscissa among the candidate coordinate points as a target coordinate point;

Calculate the reciprocal of the abscissa of the target coordinate point to obtain the resolution of the second in-film data corresponding to each of the masks, and the resolution of the second in-film data corresponding to each of the masks is formed the set of resolutions;

The resolution corresponding to each of the masks is taken as the z-axis, and the density isosurface threshold parameter of the three-dimensional density map corresponding to each of the masks and the outwardly expanded Gaussian layer parameter are taken as The x-axis and the y-axis form a three-dimensional surface map of the resolution set, and the three-dimensional surface map of the resolution set includes a dividing line.

Preferably, the S9 step further comprises the steps:

Calculate the mean value of the resolution values of each point on the dividing line;

Find the point where the resolution value is the closest to the mean value in each point on the boundary line as a discriminant point;

The resolution of the discriminant point is used as the global resolution value.

The present invention has the following beneficial effects due to the adoption of the above technical solutions:

The method for detecting the resolution of the three-dimensional density map of cryo-electron microscope of the present invention does not need to use the similarity of the reconstruction results of the two half data sets to estimate the resolution, does not need to use the original particle data set when calculating the resolution, and does not need to perform a projection process, The spectral signal-to-noise ratio of the three-dimensional density map can be directly calculated in the three-dimensional space, which has the advantages of simplicity, intuition and convenient operation.

Detailed ways

Preferred embodiments of the present invention are given below and described in detail, so that the functions and features of the present invention can be better understood.

A method for detecting the resolution of a three-dimensional density map of a cryo-electron microscope according to an embodiment of the present invention includes the steps of:

S1: Input a three-dimensional density map of cryo-EM and a three-dimensional density map of pure noise from the three-dimensional density map of cryo-EM.

For example: the data package whose ID code is EMD-8119 in the EMDataBank database can be called. The data package contains a real image and two half-dataset reconstruction results obtained by using only half of the data. The cryo-electron microscope 3D density map can use the real image pre-stored in the data package. Subtract the reconstructed half-data set reconstruction results to obtain a pure noise three-dimensional density map;

S2: Generate a mask set according to the three-dimensional density map of the cryo-electron microscope. The mask set includes n masks of different sizes, where n is an integer greater than zero.

The mask adopts a binary mask, and the binary mask includes a mask area, the value in the mask area is 1, and the value outside the mask area is 0.

In step S2, the three-dimensional cryo-EM is processed by setting and adjusting the density isosurface threshold parameter (threshold) and an outward-extended Gaussian layer parameter (nshells) of a three-dimensional density map in a transmission electron microscope image processing software EMAN2. Density map, get mask set.

When the value of the threshold parameter of the density isosurface of the 3D density map is larger or the value of the parameter of the number of Gaussian layers expanded outward is smaller, the generated mask size is smaller. By using chimera software to observe the three-dimensional density map of cryo-electron microscope, the upper and lower limits of the threshold parameters of the density isosurface of the three-dimensional density map can be obtained when a reasonable mask is generated. The lower limit of the density isosurface threshold parameter of the density map. When almost no noise is seen, the corresponding value is the upper limit of the density isosurface threshold parameter of the three-dimensional density map. The step size is generally selected so that there are ten values between the upper and lower limits. the size of the value. The number of Gaussian layers for the outward expansion is generally 2 to 11, and the step size is 1.

S3: Use each mask to segment the pure noise three-dimensional density map respectively to obtain a first in-film data and a first out-of-film data corresponding to each mask of the pure noise three-dimensional density map.

Wherein, the S3 step further comprises the steps:

Multiply the value of the pure noise three-dimensional density map on each coordinate in the three-dimensional space with the value on the corresponding coordinate of the current mask to obtain the first in-film data corresponding to the current mask;

Subtract the first in-film data from the pure noise three-dimensional density map to obtain the first out-of-film data corresponding to the current mask;

The steps are repeated until the first in-film data and the first out-of-film data corresponding to all masks are obtained.

S4: Calculate a first in-film noise spectral power and a first out-of-film noise spectral power corresponding to each mask according to the first in-film data and the first out-of-film data.

Wherein, the S4 step further comprises the steps:

Convert the first in-film data corresponding to the current mask to a first frequency space by Fourier transform; Convert the first out-of-film data corresponding to the current mask to a second frequency space by Fourier transform;

A spherical shell model shell(s) is established in the first frequency space and the second frequency space respectively, where s is the radius of the spherical shell model;

Calculate the first in-film noise spectral power and the first out-of-film noise spectral power corresponding to the current mask according to a formula (2):

Among them, RPS(s) is the spectral power; when RPS(s) is the noise spectral power in the first membrane, K _s is the coordinates of the point on the spherical shell model in the first frequency space, and N _s is the first frequency space in the the number of points on the spherical shell model, M(K _s ) is the Fourier transform of the data in the first membrane;

When RPS(s) is the noise spectral power outside the first membrane, K _s is the coordinates of the point on the spherical shell model in the second frequency space, N _s is the number of points on the spherical shell model in the second frequency space, M (K _s ) is the Fourier transform of the first out-of-film data.

S5: Calculate and obtain a linear parameter set λ _p according to the first in-film noise spectral power and the first out-of-film noise spectral power of each group;

where 0≤p≤n, and p is an integer, is the first in-film noise spectral power corresponding to the p-th mask, is the first out-of-film noise spectral power corresponding to the p-th mask; λ _p is the linear fitting parameter.

S6: using each mask to segment the cryo-EM three-dimensional density map respectively to obtain a second intra-membrane data and a second extra-membrane data corresponding to each mask of the cryo-EM three-dimensional density map;

S7: Calculate and obtain the second in-film data spectrum signal-to-noise ratio curve graph corresponding to each mask according to the second in-film data, the second out-of-film data and the linear parameter set.

Wherein, the S7 step further includes the steps:

S71: Convert the second in-film data corresponding to the current mask to a third frequency space by Fourier transform; convert the second out-of-film data corresponding to the current mask to a fourth frequency by Fourier transform space;

S72: Establish a spherical shell model shell(s) in the third frequency space and the fourth frequency space respectively, where s is the radius of the spherical shell model;

S73: Calculate the spectral power of the sum of the second in-film signal and noise and the second out-of-film noise spectral power corresponding to the current mask according to formula (2):

When RPS(s) is the spectral power of the sum of the signal and noise in the second membrane, K _s is the coordinates of the point on the spherical shell model in the third frequency space, and N _s is the point on the spherical shell model in the third frequency space. the number of points, M(K _s ) is the Fourier transform of the data in the second film;

When RPS(s) is the noise spectral power outside the second membrane, K _s is the coordinates of the point on the spherical shell model in the fourth frequency space, N _s is the number of points on the spherical shell model in the fourth frequency space, M (K _s ) is the Fourier transform of the second out-of-film data;

S74: Using a formula (3), the linear parameter set λ _p and the second out-of-film noise spectral power to estimate a second in-film noise spectral power of the cryo-EM three-dimensional density map:

RPS _np (s) = λ _p · RPS _op (s) (3);

Wherein, RPS _np (s) is the second in-film noise spectral power corresponding to the p-th mask; RPS _op (s) is the second out-of-film noise spectral power corresponding to the p-th mask;

S75 : subtract the spectral power of the second in-film noise from the summed spectral power of the second in-film signal and noise, and calculate to obtain a second in-film signal spectral power RPS _sp (s) of the three-dimensional density map of the cryo-electron microscope:

RPS _sp (s) = RPS _ip (s) - RPS _np (s) (4);

Wherein, RPS _ip (s) is the spectral power of the sum of the signal and noise in the second film corresponding to the p-th mask; RPS _np (s) is the second in-film noise spectral power corresponding to the p-th mask;

S76: Calculate the signal-to-noise ratio of the second in-film data spectrum according to the second in-film signal spectral power and the second in-film noise spectral power, and draw a second in-film data spectrum signal-to-noise ratio curve corresponding to the current mask ;

Specifically, the logarithm log ₁₀ SSNR of the data spectrum signal-to-noise ratio in the second film is taken to obtain the ordinate of the signal-to-noise ratio curve of the second film data spectrum, and the sphere corresponding to the signal-to-noise ratio of the data spectrum in the second film is obtained. The radius of the shell model is used as the abscissa of the signal-to-noise ratio curve of the data spectrum in the second film to form a signal-to-noise ratio curve of the data spectrum in the second film;

S77 : Repeat steps S71 to S76 until the signal-to-noise ratio curves of the second in-film data spectra corresponding to all the masks are obtained.

The signal-to-noise ratio of the data spectrum in the second film satisfies the formula (5):

Wherein, SSNR _p (s) is the signal-to-noise ratio of the second in-film data spectrum corresponding to the p-th mask.

S8: Obtain a resolution set according to the second in-film data spectrum signal-to-noise ratio curve, and draw a three-dimensional surface graph of the resolution set according to the resolution set.

Wherein, the S8 step further includes the steps:

Set a threshold, in this embodiment, set the threshold log ₁₀ 0.334≈-0.48;

At least one coordinate point to be selected whose ordinate is equal to the threshold value on the signal-to-noise ratio curve of each second in-film data spectrum is intercepted in each second in-film data spectrum signal-to-noise ratio curve, and selected among the candidate coordinate points A candidate coordinate point with the smallest abscissa is used as a target coordinate point;

Calculate the reciprocal of the abscissa of the target coordinate point to obtain the resolution of the data in the second film corresponding to each mask, and the resolution of the data in the second film corresponding to each mask forms a resolution set;

The resolution corresponding to each mask is taken as the z-axis, and the density isosurface threshold parameter of the three-dimensional density map corresponding to each mask and the number of Gaussian layers that expand outward are taken as the x-axis and y-axis, respectively, to form a resolution set. 3D Surface Map, Resolution Set The 3D surface map includes a dividing line.

S9: Calculate and obtain a global resolution value of the three-dimensional density map of the cryo-electron microscope according to the three-dimensional surface map of the resolution set.

Preferably, the S9 step further comprises the steps:

Calculate the mean of the resolution values of each point on the dividing line;

Find the point where the resolution value is closest to the mean value among the points on the dividing line as the discriminant point;

Take the resolution of the discriminant point as the global resolution value.

The present invention has been described in detail above with reference to the embodiments, and those of ordinary skill in the art can make various modifications to the present invention according to the above description. Therefore, some details in the embodiments should not be construed to limit the present invention, and the present invention will take the scope defined by the appended claims as the protection scope of the present invention.

Claims (10)

1. A cryo-electron microscope three-dimensional density map resolution detection method, comprising the steps: S1: input a three-dimensional density map of cryo-electron microscope and a three-dimensional density map of pure noise of the three-dimensional density map of cryo-electron microscope; S2: generate a mask set according to the three-dimensional density map of the cryo-electron microscope, where the mask set includes n masks of different sizes, where n is an integer greater than zero; S3: Segment the pure noise three-dimensional density map by using each of the masks to obtain a first in-film data and a first out-of-film data corresponding to each of the masks of the pure noise three-dimensional density map, respectively. ; S4: Calculate a first in-film noise spectral power and a first out-of-film noise spectral power corresponding to each of the masks according to the first in-film data and the first out-of-film data; S5: Calculate and obtain a linear parameter set λ _p according to the first in-film noise spectral power and the first out-of-film noise spectral power in each group; where 0≤p≤n, and p is an integer, is the first in-film noise spectral power corresponding to the p-th mask, is the first out-of-film noise spectral power corresponding to the p-th mask; λ _p is the linear fitting parameter; S6: Segment the cryo-EM 3D density map by using each of the masks to obtain a second intra-membrane data and a second extra-membrane data corresponding to each mask of the cryo-EM 3-D density map, respectively. ; S7: Calculate and obtain a spectral signal-to-noise ratio curve of the second in-film data corresponding to each mask according to the second in-film data, the second out-of-film data and the linear parameter set; S8: obtaining a resolution set according to the second in-film data spectrum signal-to-noise ratio curve, and drawing a three-dimensional surface graph of the resolution set according to the resolution set; S9: Calculate and obtain a global resolution value of the three-dimensional density map of the cryo-electron microscope according to the three-dimensional surface map of the resolution set. 2. The method for detecting the resolution of a three-dimensional density map of cryo-electron microscopy according to claim 1, wherein the mask adopts a binary mask, and the binary mask comprises a mask area, and the mask area The value inside is 1, and the value outside the mask area is 0. 3. Cryo-EM three-dimensional density map resolution detection method according to claim 2, is characterized in that, in described S2 step, by setting and adjusting the density of a three-dimensional density map etc. in a transmission electron microscope image processing software The mask set is obtained by processing the three-dimensional density map of the cryo-electron microscope using a surface threshold parameter and an outward-extended Gaussian layer parameter. 4. Cryo-EM three-dimensional density map resolution detection method according to claim 3, is characterized in that, described S3 step further comprises the step: Multiplying the value of the pure noise three-dimensional density map on each coordinate in the three-dimensional space by the value on the corresponding coordinate of the current mask to obtain the first in-film data corresponding to the current mask; subtracting the first in-film data from the pure noise three-dimensional density map to obtain the first out-of-film data corresponding to the current mask; The steps are repeated until the first in-film data and the first out-of-film data corresponding to all the masks are obtained. 5. cryo-electron microscope three-dimensional density map resolution detection method according to claim 4, is characterized in that, described S4 step further comprises the step: Converting the first in-film data corresponding to the current mask to a first frequency space through Fourier transform; converting the first out-of-film data corresponding to the current mask through Fourier transform convert to a second frequency space; A spherical shell model shell(s) is established in the first frequency space and the second frequency space respectively, where s is the radius of the spherical shell model; Calculate the first in-film noise spectral power and the first out-of-film noise spectral power corresponding to the current mask according to a formula (2): Wherein, RPS(s) is the spectral power; when RPS(s) is the spectral power of the noise in the first membrane, K _s is the coordinates of the point on the spherical shell model in the first frequency space, N _s is the number of points on the spherical shell model in the first frequency space, and M(K _s ) is the Fourier transform of the data in the first membrane; When RPS(s) is the spectral power of the first out-of-membrane noise, K _s is the coordinates of the point on the spherical shell model in the second frequency space, and N _s is the The number of points on the spherical shell model, M(K _s ) is the Fourier transform of the first out-of-membrane data. 6. Cryo-EM three-dimensional density map resolution detection method according to claim 5, is characterized in that, described S7 step further comprises the step: S71: Convert the second in-film data corresponding to the current mask to a third frequency space through Fourier transform; convert the second out-of-film data corresponding to the current mask through Fourier transform The leaf transform transforms into a fourth frequency space; S72: Establish a spherical shell model shell(s) in the third frequency space and the fourth frequency space respectively, where s is the radius of the spherical shell model; S73: Calculate the spectral power of the sum of a second in-film signal and noise and a second out-of-film noise spectral power corresponding to the current mask according to the formula (2): When RPS(s) is the spectral power of the sum of the signal and noise in the second membrane, K _s is the coordinates of the point on the spherical shell model in the third frequency space, and N _s is the third frequency space. the number of points on the spherical shell model in the frequency space, M(K _s ) is the Fourier transform of the data in the second membrane; When RPS(s) is the spectral power of the second out-of-membrane noise, K _s is the coordinates of the point on the spherical shell model in the fourth frequency space, and N _s is the the number of points on the spherical shell model, M(K _s ) is the Fourier transform of the second out-of-membrane data; S74: Using a formula (3), the linear parameter set λ _p and the second out-of-film noise spectral power to estimate a second in-film noise spectral power of the cryo-EM three-dimensional density map: RPS _np (s) = λ _p · RPS _op (s) (3); Wherein, RPS _np (s) is the second in-film noise spectral power corresponding to the pth mask; RPS _op (s) is the second out-of-film noise spectral power corresponding to the pth mask; S75: Calculate the spectral power of the second in-film signal by subtracting the spectral power of the second in-film noise from the summed spectral power of the second in-film signal and noise to obtain a second in-film signal spectral power RPS _sp (s ): RPS _sp (s) = RPS _ip (s) - RPS _np (s) (4); Wherein, RPS _ip (s) is the spectral power of the sum of the second in-film signal and noise corresponding to the p-th mask; RPS _np (s) is the second in-film noise corresponding to the p-th mask spectral power; S76: Calculate a signal-to-noise ratio of a second in-film data spectrum according to the second in-film signal spectral power and the second in-film noise spectral power, and draw the second in-film data spectral signal corresponding to the current mask Noise ratio curve; S77: Repeat steps S71-S76 until the second in-film data spectrum signal-to-noise ratio curves corresponding to all the masks are obtained. 7. The three-dimensional density map resolution detection method of cryo-electron microscope according to claim 6, is characterized in that, the signal-to-noise ratio of the data spectrum in the second film satisfies formula (5): Wherein, SSNR _p (s) is the signal-to-noise ratio of the second in-film data spectrum corresponding to the p-th mask. 8. The method for detecting three-dimensional density map resolution of cryo-electron microscope according to claim 7, characterized in that, in the step of drawing the second in-film data spectrum signal-to-noise ratio curve graph corresponding to the current mask: The logarithm of the signal-to-noise ratio of the data spectrum in the second film is taken to obtain the ordinate of the signal-to-noise ratio curve of the data spectrum in the second film, and the sphere corresponding to the signal-to-noise ratio of the data spectrum in the second film is calculated The radius of the shell model is used as the abscissa of the second in-film data spectrum signal-to-noise ratio curve graph to form the second in-film data spectrum signal-to-noise ratio curve graph. 9. Cryo-EM three-dimensional density map resolution detection method according to claim 8, is characterized in that, described S8 step further comprises step: set a threshold; At least one coordinate point to be selected whose ordinate is equal to the threshold value on the signal-to-noise ratio curve of each second in-film data spectrum is intercepted in each of the second in-film data spectrum signal-to-noise ratio curves, and Selecting the candidate coordinate point with the smallest abscissa among the candidate coordinate points as a target coordinate point; Calculate the reciprocal of the abscissa of the target coordinate point to obtain the resolution of the second in-film data corresponding to each of the masks, and the resolution of the second in-film data corresponding to each of the masks is formed the set of resolutions; The resolution corresponding to each of the masks is taken as the z-axis, and the density isosurface threshold parameter of the three-dimensional density map corresponding to each of the masks and the outwardly expanded Gaussian layer parameter are taken as The x-axis and the y-axis form a three-dimensional surface map of the resolution set, and the three-dimensional surface map of the resolution set includes a dividing line. 10. Cryo-EM three-dimensional density map resolution detection method according to claim 9, is characterized in that, described S9 step further comprises step: Calculate the mean value of the resolution values of each point on the dividing line; Find the point where the resolution value is the closest to the mean value in each point on the boundary line as a discriminant point; The resolution of the discriminant point is used as the global resolution value.