A protein ensemble prediction method using asymptotic search MSA

Abstract

A protein ensemble prediction method using asymptotic search MSA, first, given the target protein sequence, MMSeqs is used to generate MSA. Then, DBSCAN method is used for clustering each MSA to generate multiple sub-MSA; for the MSA that fails to cluster, Gaussian mixture model (GMM) is used for expectation maximization (EM) clustering; then, the BLAST sequence alignment method is used to select the most similar sequence from each sub-MSA, and MMSeqs is used again to generate a new MSA, all sub-MSA is combined into a new MSA, the same sequence is removed, and MAFFT is used to add GAP to the sequence with different lengths; the processed MSA is clustered again, and finally AlphaFold2 is used to generate the protein ensemble. The present application effectively utilizes the MSA coevolution information to predict the protein dynamic ensemble.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bioinformatics and computer application, and particularly relates to a protein ensemble prediction method using asymptotic search MSA. BACKGROUND

[0002] Proteins, as the main executors of life activities, are the components of cells and have a wide range of functions, playing a crucial role in almost all biological processes. However, proteins should not be regarded as static single structures, but as a conformational ensemble containing various accessible states, which reveal the relationship between proteins and ligand binding, enzyme catalysis, signal transduction and other biological processes. Therefore, obtaining the conformational ensemble of proteins is crucial for elucidating protein function.

[0003] Among the current methods for predicting protein ensembles, traditional experimental methods such as X-ray crystallography and nuclear magnetic resonance (NMR) can provide static structural information of proteins, but usually cannot capture dynamic behavior. In recent years, molecular dynamics simulation (MD) has been applied to dynamic structure prediction. This method can explore the movement of proteins under different conditions and capture their dynamic changes, but it has problems such as high computational cost and short time scale. With the rapid development of machine learning, methods combining machine learning and MD to predict protein ensembles have also made some achievements, such as idpGAN, ProMD, etc., but these methods also face the problems mentioned above. In today's protein structure prediction, AlphaFold2 has made significant progress in static protein structure prediction, providing a solid foundation for many protein structure prediction work, but the results predicted by AlphaFold2 are the final stable static structures, and do not consider the dynamic capture of protein structures. Therefore, many methods have been developed to modify the AlphaFold2 pipeline to predict protein dynamic structures, such as AF-Cluster, SPEACH_AF, AFsample2, AlphaFLOW, etc., but among these methods, only AlphaFLOW predicts protein ensembles, and the others only predict multiple conformations of proteins. AlphaFLOW also has problems such as high computational resource demand and slow speed. Therefore, how to maintain efficiency in predicting protein ensembles under limited resources has been a major challenge in biology. SUMMARY

[0004] In order to overcome the shortcomings of the prior art, in order to effectively utilize the MSA co-evolution information to predict the protein dynamic ensemble, the present application proposes a protein ensemble prediction method using asymptotic search MSA, a plurality of clusters are obtained by clustering the MSA searched by the current protein sequence, the cluster center in each cluster, i.e. the sequence with the highest sequence similarity score, is obtained, the corresponding MSA is searched using the sequence with the highest score, the above operation is performed again to obtain new MSA, all obtained MSAs are combined and clustered, finally, each cluster obtained by clustering is input into AlphaFold2 to predict the protein ensemble.

[0005] The technical scheme adopted by the present application to solve its technical problems is:

[0006] A protein ensemble prediction method using asymptotic search MSA, the method comprises the following steps:

[0007] 1) Given the sequence information of the target protein;

[0008] 2) According to the given sequence information of the target protein, use MMSeqs to generate multiple sequence alignment MSA;

[0009] 3) For the generated MSA, clustering is performed, and the sequences with a GAP proportion greater than a set threshold are removed during clustering;

[0010] 4) For the sub-MSA after clustering, use the BLAST sequence alignment method to find the sequence center;

[0011] 5) Use the sequence center in the sub-MSA to generate the corresponding MSA again using MMSeqs;

[0012] 6) For the new set of MSAs generated, repeat the processing procedures of steps 2) to 5) above;

[0013] 7) Combine the multiple sub-MSAs obtained by circulation into a new MSA, and ensure that each initial sequence corresponds to only one integrated MSA;

[0014] 8) Remove duplicate sequences in database search;

[0015] 9) For sequences with different sequence lengths and initial sequences in the MSA, use the MAFFT multiple sequence alignment tool to insert appropriate GAPs;

[0016] 10) Cluster using the method of step 3), and retain sequences with a GAP proportion greater than a set threshold;

[0017] 11) Input the processed multiple MSAs into AlphaFold2 to generate protein conformational ensemble.

[0018] Further, the process of 3) is as follows:

[0019] 3.1) MSA clustering using DBSCAN to get multiple sub-MSAs. DBSCAN is a commonly used density-based clustering algorithm. It discovers clusters in data by density and can identify noise. The two core conditions for DBSCAN clustering determination are epsilon (epsilon) and minPts, which represent the maximum distance of the neighborhood and the minimum number of points in the epsilon neighborhood;

[0020] 3.2) For MSA that cannot be successfully clustered by DBSCAN clustering method, use the expectation maximization EM clustering of Gaussian mixture model GMM.

[0021] Preferably, the process of 3.2) is as follows:

[0022] 3.2.1) Calculate the probability of each data point x i belonging to the kth cluster:

[0023]

[0024] where γ ik represents the posterior probability of data point x i belonging to the kth cluster, which measures the possibility of each cluster k explaining the Gaussian distribution of data point x i . π k represents the prior probability of each Gaussian distribution, i.e. the proportion of data points belonging to the cluster. N is the probability density function of Gaussian distribution, μ k is the mean vector of the kth Gaussian distribution, ∑k is the covariance matrix of the kth Gaussian distribution, N(x i |μ k ,∑k) is the probability density function value of x i on the kth Gaussian distribution. represents the total probability density of all x i explained by all K Gaussian distributions;

[0025] 3.2.2) Update parameters: use the probability calculated in step 3.2.1) to update the mean, covariance and weight of each cluster:

[0026] Mean:

[0027]

[0028] where μ k is the mean vector of the kth Gaussian distribution. Weighted sum of all data points x i , the weight is the posterior probability of its belonging to the kth cluster.

[0029] Covariance:

[0030]

[0031] where∑k is the covariance matrix of the k-th Gaussian distribution. is the weighted covariance sum over all data points, with weights being their posterior probability of belonging to the k-th cluster.

[0032] Weights:

[0033]

[0034] whereπ k is the weight of the k-th Gaussian component. N is the total number of samples, used for normalization.

[0035] 3.2.3) Repeat step 3.2.1) and step 3.2.2) until the parameters of the model, i.e. the means, covariances, weights, converge, i.e. the changes of the parameters are very small, or a preset number of iterations is reached.

[0036] Further, the process of 4) is as follows:

[0037] 4.1) Run BLASTP for alignment for each pair of sequences, and calculate their Bitscore;

[0038] 4.2) Summarize the alignment score of each query sequence, and calculate its average alignment score with other sequences;

[0039] 4.3) The one with the highest score is regarded as the sequence center.

[0040] Preferably, the process of 4.1) is as follows:

[0041] 4.1.1) BLASTP is a tool for protein sequence alignment, which is part of the BLAST tool family. It finds local similarities by comparing the input protein sequence with the protein sequences in the database, and calculates their alignment scores;

[0042] 4.1.2) Bitscore is an important scoring index in BLAST alignment results, which is used to measure the similarity between two sequences;

[0043]

[0044] where S ′is the normalized alignment score, which measures the significance of the alignment between two sequences. S is the raw alignment score, which reflects the quality of the alignment between two sequences. Lambda is the statistical scaling factor, which makes the alignment score have a uniform scale. K is the statistical correction factor, which standardizes the significance of the alignment, so that the alignment results have comparability in different backgrounds.

[0045] Furthermore, the process of 9) is as follows:

[0046] 9.1) First, read all the input A3M files and extract the sequences;

[0047] 9.2) Then, use MAFFT to align these adjusted length sequences. MAFFT will automatically add necessary GAPs during alignment according to the differences between sequences. If there are gaps between sequences at some positions, MAFFT will insert GAPs at these positions so that all sequences can be aligned at these positions;

[0048] 9.3) The alignment result output by MAFFT will contain inserted GAPs, which represent the absence of corresponding amino acids or nucleotides at that position. The aligned sequences will be written into the output file, ensuring that all sequences have consistent lengths and are aligned.

[0049] Preferably, in 3) and 10), the threshold is set to 25%.

[0050] The technical idea of the present application is as follows: first, given a target protein sequence, generate an MSA using MMSeqs. Then, use the DBSCAN method to cluster each MSA to generate multiple sub-MSAs; for the MSA that fails to cluster, use the Gaussian Mixture Model (GMM) for Expectation Maximization (EM) clustering. Next, use the BLAST sequence alignment method to select the most similar sequences from each sub-MSA, and use MMSeqs again to generate a new MSA. Combine all sub-MSAs into a new MSA, remove identical sequences, and use MAFFT to add GAPs to sequences of different lengths. Process the MSA again, and finally input AlphaFold2 to generate a protein ensemble. The present application provides a more diverse protein conformation ensemble.

[0051] The beneficial effects of the present application are: first, the MSA obtained by the first sequence search retains basic coevolution information; second, by searching the MSA with the highest sequence alignment score, and combining it with the original MSA to form a new MSA, a variety of coevolution information is added. Overall, both the initial coevolution information and the coevolution information associated with the initial coevolution information are retained. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is the whole flow chart of the method.

[0053] Figure 2 is the protein dynamic ensemble predicted by the protein ensemble prediction method with asymptotic search MSA for 6uof_A protein. DETAILED DESCRIPTION

[0054] The application will be further described below with reference to the accompanying drawings.

[0055] Reference Figure 1 and Figure 2 A protein ensemble prediction method with asymptotic search MSA, the method comprising the following steps:

[0056] 1) Given the sequence information of the target protein;

[0057] 2) According to the given sequence information of the target protein, generate a multiple sequence alignment MSA using MMSeqs;

[0058] 3) For the generated MSA, perform clustering, and remove sequences with a GAP proportion greater than a set threshold (25%) in the sequence during clustering, the process is as follows;

[0059] 3.1) Use DBSCAN to cluster the MSA to obtain multiple sub-MSAs. DBSCAN is a commonly used density-based clustering algorithm. It discovers clusters in data by density and can identify noise. The two core conditions for DBSCAN clustering determination are epsilon and minPts, which mainly represent the maximum distance of the neighborhood and the minimum number of points in the epsilon neighborhood;

[0060] 3.2) For the MSA that cannot be successfully clustered by the DBSCAN clustering method, use the expectation maximization EM clustering of Gaussian mixture model GMM; the process is as follows:

[0061] 3.2.1) Calculate the probability of each data point x i belonging to the kth cluster:

[0062]

[0063] where γ ik represents the posterior probability of the kth cluster of data point x i , which measures the possibility of each cluster k explaining the Gaussian distribution of data point x i . π k represents the prior probability of each Gaussian distribution, that is, the proportion of data points belonging to the cluster. N is the probability density function of Gaussian distribution, μ k is the mean vector of the kth Gaussian distribution, ∑k is the covariance matrix of the kth Gaussian distribution, and N(xi |μ k ,∑k) is the probability density function value on the kth Gaussian distribution. i represents all x i The total probability density explained by all K Gaussian distributions.

[0064] 3.2.2) Update parameters: using the probabilities calculated in step 3.2.1), update the mean, covariance and weight of each cluster:

[0065] Mean:

[0066]

[0067] where μ k is the mean vector of the kth Gaussian distribution. is the weighted sum of all data points x i with the weight being its posterior probability of belonging to the kth cluster.

[0068] Covariance:

[0069]

[0070] where ∑k is the covariance matrix of the kth Gaussian distribution. is the weighted covariance sum of all data points with the weight being its posterior probability of belonging to the kth cluster.

[0071] Weight:

[0072]

[0073] where π k is the weight of the kth Gaussian component. N is the total number of samples for normalization.

[0074] 3.2.3) Repeat steps 3.2.1) and 3.2.2) until the parameters (mean, covariance, weight) of the model converge, i.e., the change of parameters is very small, or reach a preset number of iterations.

[0075] 4) For the sub-MSA after clustering is completed, use the BLAST sequence alignment method to find the sequence center; the process is as follows:

[0076] 4.1) Run BLASTP for each pair of sequences (such as sequence 1 and sequence 2, sequence 1 and sequence 3, etc.) for alignment and calculate their Bitscore; the process is as follows:

[0077] ​4.1.1) BLASTP is a tool for protein sequence alignment, part of the BLAST tool family, which finds local similarities between input protein sequences and protein sequences in the database, and calculates their alignment scores;

[0078] 4.1.2) Bitscore is an important scoring indicator in BLAST alignment results, used to measure the similarity between two sequences;

[0079]

[0080] where S ′ is the normalized alignment score, used to measure the significance of the alignment between two sequences.

[0081] S is the original alignment score, reflecting the quality of the alignment between two sequences. λ is the statistical proportionality coefficient, so that the alignment score has a uniform scale. K is the statistical correction factor, which normalizes the significance of the alignment, so that the alignment results have comparability in different backgrounds.

[0082] 4.2) Summarize the alignment scores of each query sequence, and calculate the average alignment score with other sequences;

[0083] 4.3) The highest score is considered as the sequence center;

[0084] 5) Use the sequence center in the sub-MSA to generate the corresponding MSA using MMSeqs again;

[0085] 6) Repeat the processing steps 2) to 5) for the new set of MSAs generated in the above processing;

[0086] 7) Combine the multiple sub-MSAs obtained by the loop into a new MSA, ensuring that each initial sequence corresponds to only one integrated MSA;

[0087] 8) Remove duplicate sequences in the database search to reduce redundancy;

[0088] 9) For sequences in the MSA that have different lengths and initial sequences, use the MAFFT multiple sequence alignment tool to insert appropriate GAPs; the process is as follows:

[0089] 9.1) First read all the input A3M files and extract the sequences;

[0090] 9.2) Then, use MAFFT to align these adjusted length sequences. MAFFT will automatically add necessary GAPs during alignment according to the differences between sequences, for example, if there are deletions (such as insertions or deletions) between sequences at some positions, MAFFT will insert GAPs at these positions so that all sequences can be aligned at these positions;

[0091] 9.3) The alignment results output by MAFFT will include inserted GAPs, which represent sequences that do not have the corresponding amino acid or nucleotide at that position. The aligned sequences will be written to the output file to ensure that all sequences are of the same length and aligned.

[0092] 10) Cluster using the methods in steps 3.1) and 3.2), retaining sequences with a GAP percentage greater than a set threshold (25%);

[0093] 11) Input the processed MSA into AlphaFold2 one by one to generate a protein conformation ensemble.

[0094] This implementation case uses the transcriptional regulatory protein 6uof_A, with a sequence length of 119, as an example. The experimental results are as follows: Figure 2 As shown. A protein ensemble prediction method using asymptotic MSA search includes the following steps:

[0095] 1) Given the sequence information of the target protein;

[0096] 2) Based on the given sequence information of the target protein, use MMSeqs to generate multiple sequence alignment (MSA) data;

[0097] 3) For the generated MSA, perform clustering, and remove sequences with a GAP ratio greater than 25% during clustering;

[0098] The process is as follows:

[0099] 3.1) Use DBSCAN to perform MSA clustering to obtain multiple sub-MSA. DBSCAN is a commonly used density-based clustering algorithm that uses density to discover clusters in the data and can identify noise. The two core conditions for DBSCAN's clustering decision are ε (epsilon) and

[0100] minPts represents the maximum distance in the neighborhood and the minimum number of points in the ε-neighborhood;

[0101] 3.2) For MSA clusters that cannot be successfully clustered by the DBSCAN clustering method, the Expectation-Maximization (EM) clustering method of Gaussian Mixture Model (GMM) is used; the process is as follows:

[0102] 3.2.1) Calculate x for each data point i The probability of belonging to the k-th cluster:

[0103]

[0104] Where, γ ik Represents data point x iThe posterior probability of belonging to the kth cluster, which measures the likelihood of each cluster k explaining the data point x i k The prior probability of each Gaussian distribution, i.e. the proportion of data points belonging to this cluster. N is the probability density function of the Gaussian distribution, μ k is the mean vector of the kth Gaussian distribution, ∑k is the covariance matrix of the kth Gaussian distribution, N(x i | μ k , ∑k) is the value of the probability density function of x i on the kth Gaussian distribution. i The total probability density of all x i being explained by all K Gaussian distributions.

[0105] 3.2.2) Update parameters: using the probabilities calculated in step 3.2.1), update the mean, covariance and weight of each cluster:

[0106] Mean:

[0107]

[0108] where μ k is the mean vector of the kth Gaussian distribution. The weighted sum of all data points x i , with the weight being its posterior probability of belonging to the kth cluster.

[0109] Covariance:

[0110]

[0111] where ∑k is the covariance matrix of the kth Gaussian distribution. is the weighted covariance sum of all data points, with the weight being its posterior probability of belonging to the kth cluster.

[0112] Weight:

[0113]

[0114] where π k is the weight of the kth Gaussian component. N is the total number of samples, used for normalization.

[0115] 3.2.3) Repeat steps 3.2.1) and 3.2.2) until the parameters of the model, i.e. the mean, covariance,

[0116] weight, converge, i.e. the changes of the parameters are very small, or a preset number of iterations is reached.

[0117] 4) For the clustered sub-MSA, find the sequence center using BLAST sequence alignment method; the process is as follows:

[0118] 4.1) Run BLASTP for each pair of sequences (such as sequence 1 and sequence 2, sequence 1 and sequence 3, etc.) to perform alignment and calculate their Bitscore; the process is as follows:

[0119] 4.1.1) BLASTP is a tool for protein sequence alignment, which is part of the BLAST tool family. It finds local similarities between input protein sequences and protein sequences in the database and calculates their alignment scores;

[0120] 4.1.2) Bitscore is an important scoring indicator in BLAST alignment results, used to measure the similarity between two sequences;

[0121]

[0122] where S ′ is the normalized alignment score, used to measure the significance of alignment between two sequences.

[0123] S is the original alignment score, reflecting the alignment quality of two sequences. λ is the statistical proportion coefficient, making the alignment score have a unified scale. K is the statistical correction factor, which normalizes the significance of alignment, so that the alignment results have comparability in different backgrounds.

[0124] 4.2) Summarize the alignment score of each query sequence and calculate its average alignment score with other sequences;

[0125] 4.3) The highest score is considered as the sequence center;

[0126] 5) Use the sequence center in the sub-MSA to generate the corresponding MSA using MMSeqs;

[0127] 6) For the generated new MSA, repeat the processing steps 2) to 5) of the above processing procedure;

[0128] 7) Combine the multiple sub-MSA obtained by circulation into a new MSA, ensuring that each initial sequence corresponds to only one integrated MSA;

[0129] 8) Remove duplicate sequences in database search to reduce redundancy;

[0130] 9) For sequences with different sequence lengths and initial sequences in MSA, use MAFFT multiple sequence alignment tool to insert appropriate GAP; the process is as follows:

[0131] 9.1) First read all input A3M files and extract sequences;

[0132] 9.2) Then, use MAFFT to align these adjusted length sequences. MAFFT will automatically add necessary GAPs in the alignment process according to the differences between sequences, for example, if there are deletions (such as insertions or deletions) between sequences at some positions, MAFFT will insert GAPs at these positions so that all sequences can be aligned at these positions;

[0133] 9.3) The alignment result output by MAFFT will contain inserted GAPs, which represent that the sequence has no corresponding amino acids or nucleotides at this position. The aligned sequences will be written into the output file, ensuring that all sequences are of the same length and aligned;

[0134] 10) Use the clustering method of steps 3.1) and 3.2) to retain sequences with GAPs accounting for more than 25%;

[0135] 11) Input the processed multiple MSAs into AlphaFold2 one by one to generate protein conformation ensembles.

[0136] The above is the result of one example given by the present application. Obviously, the present application is not only suitable for the above-mentioned embodiments, but can be implemented in various ways without deviating from the basic spirit of the present application and without exceeding the content involved by the essential content of the present application.

Claims

1. A protein ensemble prediction method employing an asymptotic search for MSA, characterized in that, The method includes the following steps: 1) Given the sequence information of the target protein; 2) Based on the given sequence information of the target protein, use MMSeqs to generate multiple sequence alignment (MSA) data; 3) For the generated MSA, perform clustering, and remove sequences whose gap ratio is greater than a set threshold during clustering; 4) For the sub-MSA after clustering, the BLAST sequence alignment method is used to find the sequence center; 5) Use the sequence centers in the sub-MSA to generate the corresponding MSA again using MMSeqs; 6) For this newly generated MSA, repeat the processing steps 2) to 5) above; 7) Merge multiple sub-MSA obtained from the loop into a new MSA, ensuring that each initial sequence corresponds to only one integrated MSA; 8) Remove duplicate sequences from the database search; 9) For sequences in the MSA that differ in length from the initial sequence, use the MAFFT multiple sequence alignment tool to insert appropriate gaps; 10) Cluster using the method in step 3), retaining sequences with a GAP ratio greater than a set threshold; 11) Input the processed MSA into AlphaFold2 one by one to generate a protein conformation ensemble.

2. The protein ensemble prediction method using an asymptotic search MSA as described in claim 1, characterized in that, The process described in 3) is as follows: 3.1) DBSCAN is used to perform MSA clustering to obtain multiple sub-MSA. The two core conditions for DBSCAN clustering are ε and minPts, which represent the maximum distance of the neighborhood and the minimum number of points in the ε neighborhood. 3.2) For MSA clusters that cannot be successfully clustered by the DBSCAN clustering method, the expectation-maximization EM clustering of the Gaussian mixture model (GMM) is used.

3. The protein ensemble prediction method using an asymptotic search MSA as described in claim 2, characterized in that, The process described in 3.2) is as follows: 3.2.1) Calculate x for each data point i The probability of belonging to the k-th cluster: Where, γ ik Represents data point x i The posterior probability of belonging to the k-th cluster is used to measure the probability of each cluster k pairs of data points x. i The probability explained by the Gaussian distribution, π k Let μ represent the prior probability of each Gaussian distribution, i.e., the proportion of data points belonging to that cluster. N is the probability density function of the Gaussian distribution. k Let N(x) be the mean vector of the k-th Gaussian distribution, and let ∑k be the covariance matrix of the k-th Gaussian distribution. i |μ k ,∑k) is x i The probability density function value on the k-th Gaussian distribution. Represents all x i The total probability density explained by all K Gaussian distributions; 3.2.2) Update parameters: Using the probabilities calculated in step 3.2.1), update the mean, covariance, and weights of each cluster: Mean: Where, μ k It is the mean vector of the k-th Gaussian distribution. For all data points x i The weighted summation, where the weights are the posterior probabilities of belonging to the k-th cluster; Covariance: Where ∑k is the covariance matrix of the k-th Gaussian distribution; It is the sum of the weighted covariances of all data points, where the weights are the posterior probabilities of each data point belonging to the k-th cluster. Weight: Where, π k is the weight of the k-th Gaussian component, and N is the total number of samples, used for normalization; 3.2.3) Repeat steps 3.2.1) and 3.2.2) until the model parameters, namely the mean, covariance, and weights, converge, meaning the parameter changes are very small, or the preset number of iterations is reached.

4. A protein ensemble prediction method using an asymptotic search MSA as described in any one of claims 1 to 3, characterized in that, The process described in 4) is as follows: 4.1) Perform BLASTP on each pair of sequences for alignment and calculate their Bitscore; 4.2) Summarize the alignment scores of each query sequence and calculate its average alignment score with other sequences; 4.3) The one with the highest score is considered the sequence center.

5. The protein ensemble prediction method using an asymptotic search MSA as described in claim 4, characterized in that, The process described in 4.1) is as follows: 4.1.1) BLASTP is a tool for protein sequence alignment, part of the BLAST tool family. It finds local similarities by comparing an input protein sequence with protein sequences in a database and calculates their alignment scores. 4.1.2) Bitscore is an important scoring metric in BLAST alignment results, used to measure the similarity between two sequences; Among them, S ′ S is the standardized alignment score, used to measure the significance of the alignment between two sequences. S is the original alignment score, reflecting the alignment quality of the two sequences. λ is the statistical proportionality coefficient, which makes the alignment score have a uniform scale. K is the statistical correction factor, which standardizes the significance of the alignment, making the alignment results comparable under different backgrounds.

6. A protein ensemble prediction method using an asymptotic search MSA as described in any one of claims 1 to 3, characterized in that, The process described in 9) is as follows: 9.1) First, read all input A3M files and extract the sequences; 9.2) Then, MAFFT is used to align these adjusted length sequences. MAFFT will automatically add necessary gaps during the alignment process based on the differences between the sequences. If there are gaps between the sequences at certain positions, MAFFT will insert gaps at these positions so that all sequences can be aligned at these positions. 9.3) The alignment results output by MAFFT will include inserted GAPs, which represent sequences where the corresponding amino acid or nucleotide is absent at that position. The aligned sequences will be written to the output file. Ensure that all sequences have the same length and are aligned.

7. A protein ensemble prediction method using an asymptotic search MSA as described in any one of claims 1 to 3, characterized in that, In 3) and 10), the threshold is set to 25%.

Images