CN-122024874-A - Automatic single-molecule force spectrum signal analysis method based on optical tweezers

CN122024874ACN 122024874 ACN122024874 ACN 122024874ACN-122024874-A

Abstract

The invention discloses an automatic single-molecule force spectrum signal analysis method based on optical tweezers, and aims to solve the problems of low analysis efficiency, strong subjectivity and signal loss of a noise reduction algorithm in the prior art. The method comprises the steps of generating a simulated signal slice data set through a single-molecule stretching model, training a fusion perception self-adaptive noise reduction model, carrying out overlapped sliding slicing and normalization on force-distance signals acquired by optical tweezers, inputting the model for denoising and reconstructing the signals, and finally obtaining parameters such as the size, free energy and the like of a molecular folding structure through automatic fitting of a secondary differential automatic positioning folding site and a worm chain model and automatic parameter calculation, thereby realizing analysis and automation of single-molecule force spectrum signals acquired by the optical tweezers. The method breaks through the balance problem of noise reduction and signal fidelity, remarkably improves the efficiency, accuracy and stability of single-molecule force spectrum analysis, and is suitable for single-molecule mechanics research in the fields of biophysics and molecular biology.

Inventors

LI YANGHUI
CHEN LINYAO
ZHANG XIAOPENG
Jiang Shenhao
WANG WANYING
WANG LE

Assignees

中国计量大学

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (8)

1. An automatic single-molecule force spectrum signal analysis method based on optical tweezers is characterized by comprising the following steps: 1) Constructing a single-molecule stretching model, generating a single-molecule force spectrum simulation signal by using the model, wherein the single-molecule force spectrum simulation signal comprises a noisy and noiseless force-distance simulation signal, and carrying out sectional processing on the simulation signal to obtain a noisy and noiseless simulation signal slice data set; 2) Constructing a fusion perception self-adaptive noise reduction model suitable for processing force-distance signal slices, and training the fusion perception self-adaptive noise reduction model by utilizing the simulation signal slice data set generated in the step 1) to obtain a trained fusion perception self-adaptive noise reduction model; 3) Overlapping sliding slices are carried out on force-distance signals acquired by the optical tweezers, signal values are normalized to an [ -1,1] interval by using Max-Min normalization, normalized signal slices are obtained, the normalized signal slices are input into a fusion perception self-adaptive noise reduction model trained in the step 2), and the noise reduction signal slices are output after model processing; 4) Carrying out inverse Max-Min normalization on the denoising signal slice in the step 3), and then carrying out synthesis processing to obtain a recombined force-distance signal; 5) The automatic analysis of the single-molecule force spectrum signal is realized by sequentially carrying out automatic positioning of folding signal sites, automatic fitting of worm chain models and automatic parameter calculation on the recombined force-distance signals in the step 4).
2. The method for automatically analyzing the single-molecule force spectrum signal based on the optical tweezers according to claim 1 is characterized in that the single-molecule stretching model in the step 1) is constructed by randomly selecting a value in the range of 30-80 nanometers as a holding force length P, randomly selecting 3 different contour lengths L in the range of 500-1000 nanometers, substituting three groups of holding force lengths P and contour lengths L into a worm chain model to generate corresponding 3 curves, wherein the abscissa of the curves is a distance ordinate, taking a section of curve randomly on the 3 curves as a signal curve, and the adjacent signal curves are partially overlapped on the force value, and are not overlapped on the distance value, so as to construct an analog signal of a force-distance signal, then performing analog-to-digital conversion on the analog signal with the frequency of 100 Hz to obtain a discrete force-distance digital signal, namely a noiseless force-distance analog signal, finally, respectively adding Gaussian noise of the force signal and the distance signal to the noiseless analog signal to obtain a noiseless force-distance digital signal, namely, segmenting the noiseless force-distance analog signal, and processing the noiseless force-distance analog signal to be divided into sections according to the respective slice numbers of 300.
3. The automatic analysis method of the single-molecule force spectrum signal based on the optical tweezers of claim 1 is characterized in that the structure of the fusion perception self-adaptive noise reduction model in the step 2) comprises 6 multi-layer fusion perception modules, 2 self-attention modules and 1 output convolution layer, wherein each 3 multi-layer fusion perception modules are connected in series and then are connected with 1 self-attention module to form 2 series combined structures in a conformal mode, jump connection is added between the combined structures, finally, characteristic parameters generated by the combined structures are summarized by 1 output convolution layer to output a denoising signal slice, the input and output of the fusion perception self-adaptive noise reduction model are two-channel data, the two-channel data are respectively corresponding to a force channel and a distance channel of the stress-distance signal slice, the multi-layer fusion perception module is composed of 4 one-dimensional convolution kernels and 1 maximum pooling layer, the number of input and output channels is 80, the number of each convolution kernel channel is 16, the sizes are 7, 15, 29 and 59, the number of the input and output channels of the maximum pooling layer is 16, the number of convolution kernels is 9, the number of one-dimensional convolution kernels is 80, and the number of output channels is 1.
4. The method for automatically analyzing the single-molecule force spectrum signal based on the optical tweezers according to claim 1, wherein the training the fusion perception self-adaptive noise reduction model in the step 2) means that a noisy simulation signal slice is used as input data of the model, a noise-free simulation signal slice is used as an output supervision tag of the model, and model parameters are optimized by an Adam optimizer by taking a mean square error as a loss function, so that the trained fusion perception self-adaptive noise reduction model is finally obtained.
5. The automatic analysis method of single-molecule force spectrum signals based on optical tweezers according to claim 1, wherein in the step 3), sliding slices refer to overlapping and dividing force signals and distance signals acquired by the optical tweezers by taking 300 data points as one slice, the overlapping rate of adjacent slices is 97%, when the number of data points of the slices is less than 300, the last value of the slice is adopted to supplement the slices to 300 data points, the synthesis process refers to sequentially connecting all the slices normalized by inverse Max-Min according to the sequence of the sliding slices, and the numerical value of the overlapping part in the adjacent slices is averaged.
6. The automatic analysis method of single-molecule force spectrum signals based on optical tweezers according to claim 1, wherein the automatic positioning of folding signal sites in the step 5) means that the force signals in the force-distance signals after recombination are subjected to a secondary differential operation, the differential interval is set to 10, a secondary differential curve is obtained, then the peak value and the number of peaks larger than 0.5 threshold value in the secondary differential curve are identified, and the values of the force signals and the distance signals at the time point, namely the folding signal sites, are automatically positioned according to the time point where each peak value is located.
7. The automatic analysis method of single-molecule force spectrum signals based on optical tweezers according to claim 1, wherein the step 5) of automatic worm chain model fitting refers to segmenting the recombined force and distance signals according to folding signal sites, and then automatically performing curve fitting on each segment by adopting a worm chain model to obtain the parameters of the worm chain model, namely the holding force length and the contour length corresponding to each segment of force and distance signals.
8. The automatic analysis method of single-molecule force spectrum signals based on optical tweezers according to claim 1, wherein the automatic parameter calculation in the step 5) is to automatically calculate parameters of a folded structure of a force-distance curve, namely a single-molecule force spectrum signal, based on parameters fitted by a worm chain model, and the obtained true dimension and free energy of the folded structure and DNA handle information connected to the single molecule.

Description

Automatic single-molecule force spectrum signal analysis method based on optical tweezers Technical Field The invention relates to the field of single-molecule detection, in particular to an automatic single-molecule force spectrum signal analysis method based on optical tweezers. Background The optical tweezers single-molecule force spectrum technology captures the microsphere through an optical trap formed by a laser beam, single molecules are indirectly controlled, and the relation between molecular stress and deformation can be determined by detecting the deflection of the microsphere in the optical trap. Can be used for researching DNA folding, protein conformational change, antigen-antibody interaction and the like, and has wide application in the fields of biophysics and molecular biology. The analysis of the core signal force-distance curve signal of the single-molecule force spectrum technology depends on manual labeling and manual parameter extraction, and has the limitations of low efficiency, strong subjectivity, low data utilization rate and the like, while the automatic analysis technology realizes batch processing, feature identification and parameter quantification of the force-distance signal through an algorithm, so that the efficiency and accuracy of processing the single-molecule force spectrum are greatly enhanced. The Brownian motion makes the microsphere captured by the light trap shake randomly, so that irregular fluctuation of force-distance signals occurs, system noise is superimposed, the force-distance signals cover the real mechanical response of molecules together, the accuracy of the signals is reduced, meanwhile, the characteristic recognition of an automatic algorithm is interfered, the analysis performance is restricted, the noise reduction algorithms such as mean value filtering, kalman filtering, low-pass Butterworth filtering and local weighted regression can cause signal loss when the noise is reduced, and the stability and the accuracy of the force-distance signal analysis algorithm cannot be guaranteed in the noise reduction accuracy. Disclosure of Invention Aiming at the problems of insufficient precision and low stability under noise interference of the existing automatic force-distance signal analysis algorithm, the invention provides an automatic single-molecule force spectrum signal analysis method based on optical tweezers. The force-distance signals are subjected to sliding slice segmentation, the force-distance signals with high signal to noise ratio are obtained after denoising and slice recombination of the fusion perception self-adaptive noise reduction model, and finally, molecular structure parameters in the force-distance signals are automatically calculated by using a quadratic difference method, worm chain model fitting, free energy integration and other methods, so that automatic analysis of single-molecule force spectrum signals of the pair of optical tweezers is realized, and analysis efficiency and analysis precision are greatly improved. The invention is realized by the following technical scheme: 1) Constructing a single-molecule stretching model, generating a single-molecule force spectrum simulation signal by using the model, wherein the single-molecule force spectrum simulation signal comprises a noisy and noiseless force-distance simulation signal, and carrying out sectional processing on the simulation signal to obtain a noisy and noiseless simulation signal slice data set; Firstly, randomly selecting a value in the range of 30-80 nanometers as a holding force length P, randomly selecting 3 different profile lengths L in the range of 500-1000 nanometers, substituting three groups of holding force lengths P and profile lengths L into a worm chain model to generate 3 corresponding curves, wherein the abscissa of the curves is the distance ordinate as force; then, a section of curve is randomly intercepted on the 3 curves to be used as a signal curve, adjacent signal curves are partially overlapped on force values, and the adjacent signal curves are not overlapped on distance values to construct an analog signal of a force-distance signal, then, the analog signal is subjected to analog-to-digital conversion with the frequency of 100 Hz to obtain a discrete force-distance digital signal, namely a noiseless force-distance simulation signal, and finally, gaussian noise of the force signal and the distance signal is respectively added to the noiseless simulation signal to obtain a noisy force-distance digital signal, namely a noisy force-distance simulation signal; the analog-to-digital conversion formula is as follows: And Wherein x actual represents the actual displacement between two microbeads, x 0 is the initial distance between the two microbeads, v trap represents the set moving speed of the optical trap, t is a discrete time point sequence in the moving process of the optical tweezers, F is the instantaneous molecular tension applied betwee