Method for predicting CTCF chromatin ring based on transcription factor sequence in ring extrusion model

Abstract

The invention relates to a method for predicting a CTCF chromatin ring based on a transcription factor sequence in a ring extrusion model. The problems that in the prior art, a chromatin ring prediction method is limited in interpretability and inaccurate in prediction are solved. The method comprises the following steps: S1, constructing a data set; s2, key transcription factor sequence transformation; s3, establishing a CTCF chromatin ring classification model based on a key transcription factor sequence; and S4, performing model evaluation. The method has the advantages that only the key transcription factor sequence is used as input, and prediction of the CTCF chromatin ring can be achieved. A transcription factor combination sequence is innovatively coded into a transcription factor sequence, a pre-training BERT model of a natural language is utilized, the transcription factor sequence is decoded, prediction of a CTCF chromatin ring is achieved, the prediction accuracy is improved, and meanwhile the interpretability of a prediction model is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care informatics, and in particular to a method for predicting CTCF chromatin loops based on transcription factor sequences in a loop extrusion model. Background Art

[0002] Existing methods for predicting CTCF chromatin loops include machine learning methods (CTCF-MP, Lollipop, CCIP) and mathematical modeling methods (LC models) based on sample genome sequences and epigenomic features. Existing algorithms are all based on complex genome sequences and epigenomic features, including various DNA sequence compositions, histone modifications, gene expression RNA-seq signals, genomic binding signals of architectural proteins CTCF and RAD21, loop length, CTCFMot if direction, etc. as inputs. They simply rely on the learning ability of machine learning or deep learning technology, have black box problems, and have limited interpretability. Based on simple mathematical models, it is often necessary to manually select and construct features, such as the LC model, which may miss important information and lead to inaccurate prediction results. Summary of the invention

[0003] The purpose of the present invention is to provide a method for predicting CTCF chromatin loops based on transcription factor sequences in a loop extrusion model, in response to the current chromatin loop prediction methods with poor interpretability of characteristic patterns.

[0004] To achieve the above object, the present invention adopts the following technical solution: a method for predicting CTCF chromatin loops based on transcription factor sequences in a loop extrusion model, characterized in that the method comprises the following steps:

[0005] S1. Build a dataset.

[0006] S2, key transcription factor sequence conversion;

[0007] S3. Establish a CTCF chromatin loop classification model based on key transcription factor sequences;

[0008] S4. Model evaluation.

[0009] In step S1, the chromatin loops are detected using Hi-C experimental data, and the position coordinates of transcription factors in the genome are obtained through ChIP-seq data of transcription factors in the cell differentiation stage.

[0010] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S1 includes the following steps:

[0011] S11, obtain all CTCF combination patterns within 1 Mb;

[0012] S12, select the chromatin loops verified by Hi-C experiments in all CTCF pairs as the positive set;

[0013] S13, remove the positive set Positive Set and Bin Set trash data, and the remaining is the negative set negative_pool.

[0014] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S11 includes the following steps:

[0015] S111, firstly, all CTCFs in the SC-βorg stage of pancreatic cell differentiation were combined in pairs within each chromosome;

[0016] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S12 includes the following steps:

[0017] S121, remove all CTCF pairs outside the 1Mb range, retain the starting coordinates of the latter CTCF - the ending coordinates of the former CTCF, and call this data set 1mb_ctcf_pairs, then perform overlap analysis on all 1mb_ctcf_pairs and Hi-C data to find out the regions that are included in 1mb_ctcf_pairs and overlap in Hi-C data;

[0018] S122. Select the CTCF binding site with the highest peak value in the overlapping area and define it as the positive sample, i.e., the Positive Set;

[0019] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S13 includes the following steps:

[0020] S131, the remaining overlapping area data is put into the Bin Set trash database;

[0021] S132, remove the positive set Positive Set and Bin Set trash bin data from all 1mb_ctcf_pairs, and the remaining is the negative set negative_pool.

[0022] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S2 specifically includes the following steps:

[0023] S21. Download the CTCF motif from the JASPAR database: MA0139.1.transfac;

[0024] S22, combine the FASTA file of the CTCF genomic sequence to obtain directionality data;

[0025] S23, using the bedtools tool to integrate the narrowpeak files of the five transcription factors with the Hi-C loop data to generate a file containing transcription factor information and overlapping regions;

[0026] S24. The data were sorted and processed, and finally the arrangement of key transcription factors in the chromatin loop was constructed.

[0027] In step S3, when building the classification model, 80% of the samples are randomly selected as training sets and 20% as independent test sets. The pre-trained model of BERT-Base-Uncased is used for fine-tuning and binary classification.

[0028] In the above-mentioned method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, BERT is a natural language processing model, and BERT-Base-Uncased is a specific variant of the BERT model.

[0029] In the above-mentioned method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, model building includes the following steps:

[0030] S31. First, define a preprocessing function to tokenize the sequence and fill it with a fixed length;

[0031] S32. Then define the training parameters and evaluation indicators.

[0032] Through the above steps, a classification model of CTCF chromatin loops based on BERT-Base-Uncased was established, and the classification model was used to classify the key transcription factor sequences in chromatin loops.

[0033] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S4 specifically includes the following steps:

[0034] S41. Obtain the predicted probabilities and true labels of the LC model and BERT-Base-Uncased models to calculate the AUROC and AUPRC values.

[0035] S42, calculating logistic regression of TPR and FPR based on different thresholds;

[0036] S43. Use FPR as the x-axis and TPR as the y-axis to draw the ROC curve;

[0037] S44. Measure the performance of the classifier and determine the accuracy of the classification results.

[0038] In step S41, the formula for calculating AUROC and AUPRC values ​​is:

[0039]

[0040] Among them, TP, FP, TN and FN represent the number of true positives, false positives, true negatives and false negatives respectively. TPR is the true positive rate, that is, the proportion of positive samples correctly identified as positive to all positive samples; FPR is the false positive rate, that is, the proportion of negative samples incorrectly identified as positive to all negative samples.

[0041] Compared with the prior art, the advantages of the present invention are:

[0042] 1. Using only key transcription factor sequences as input, the prediction of CTCF chromatin loops can be achieved. This can be extended to other chromatin structure and transcription factor research fields, providing new tools and ideas for genomics and epigenomics research.

[0043] 2. Through innovative feature representation methods combined with fine-tuning of natural language pre-trained models, the accuracy of CTCF chromatin loop prediction can be significantly improved.

[0044] 3. Innovatively encode the transcription factor binding order as a transcription factor sequence, use the natural language pre-trained BERT model to decode the transcription factor sequence, and realize the prediction of CTCF chromatin loops, which improves the prediction accuracy and enhances the interpretability of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a schematic diagram of a ring extrusion model in the present invention;

[0046] Figure 2 It is a data processing flow chart in the present invention;

[0047] Figure 3 It is a schematic diagram of the test results of the Bert-Base-Uncased model parameters fine-tuned by the training set in the present invention on an independent test set; DETAILED DESCRIPTION

[0048] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model comprises the following steps:

[0050] S1. Build a dataset.

[0051] S2, key transcription factor sequence conversion;

[0052] S3. Establish a CTCF chromatin loop classification model based on key transcription factor sequences;

[0053] S4. Model evaluation.

[0054] In step S1, chromatin loops are detected using Hi-C experimental data, and the position coordinates of transcription factors in the genome are obtained through ChIP-seq data of transcription factors in the SC-βorg stage of pancreatic cell differentiation.

[0055] The loop extrusion model is a widely accepted model for explaining the molecular mechanism of cohesin-mediated chromatin interactions. According to this model, specific protein complexes, such as CTCF and cohesin, bind to specific sites on chromatin and actively squeeze chromatin to form a loop structure. Finally, WAPL is responsible for releasing cohesin from chromatin, thus ending the formation of chromatin loops.

[0056] like Figure 1 As shown, according to the loop extrusion model, the chromatin architecture proteins CTCF and cohes in bind to specific sites in the genome and actively squeeze chromatin to form a ring structure with the participation of ATP. Finally, the unloading protein WAPL is responsible for releasing cohes in from chromatin, ending the extrusion of the chromatin loop.

[0057] Data processing flow Figure 2 As shown:

[0058] a Combine CTCF pairs to form a CTCF pairs dataset;

[0059] b Obtain all CTCF combination patterns within 1Mb;

[0060] c The distribution of positive and negative pools;

[0061] d Source of key transcription factor sequences: Hi-C loop and narrowpeak files of key transcription factors.

[0062] After the positive and negative set (imbalanced data set) sequences are input into the Bert-Base-Uncased model, AUROC and AUPRC are used as evaluation indicators.

[0063] The construction of positive and negative sets in step S1 data set includes the following steps:

[0064] S11, obtain all CTCF combination patterns within 1 Mb;

[0065] S12, select the chromatin loops verified by Hi-C experiments in all CTCF pairs as the positive set;

[0066] S13, remove the positive set Positive Set and Bin Set trash data, and the remaining is the negative set negative_pool.

[0067] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S11 includes the following steps:

[0068] S111, firstly, all CTCFs in the SC-βorg stage of pancreatic cell differentiation were combined in pairs within each chromosome;

[0069] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S12 includes the following steps:

[0070] S121, remove all CTCF pairs outside the 1Mb range, retain the starting coordinates of the latter CTCF - the ending coordinates of the former CTCF, and call this data set 1mb_ctcf_pairs, then perform overlap analysis on all 1mb_ctcf_pairs and Hi-C data to find out the regions that are included in 1mb_ctcf_pairs and overlap in Hi-C data;

[0071] S122. Select the CTCF binding site with the highest peak value in the overlapping area and define it as the positive sample, i.e., the Positive Set;

[0072] In the above method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model, step S13 includes the following steps:

[0073] S131, the remaining overlapping area data is put into the Bin Set trash database;

[0074] S132. Remove the positive set and bin set trash data from all 1mb_ctcf_pairs, and the remaining is the negative set pool Negat ive_pool.

[0075] The sample size of the above dataset is shown in Table 1:

[0076]

[0077] Table 1 - Sample size summary

[0078] As shown in Table 2, each transcription factor is represented by a corresponding letter, where CTCF+ and CTCF- represent the forward and reverse directions of the CTCF binding site. This directionality plays an important role in the formation of chromatin loops and affects the three-dimensional structure of the genome and the gene regulatory network.

[0079]

[0080] Table 2 - Alphabetical representation of key transcription factors

[0081] The operation of using the Strom tool to obtain the direction of CTCF is as follows:

[0082] S21. Download the CTCF motif from the JASPAR database: MA0139.1.transfac;

[0083] S22, combine the FASTA file of the CTCF genomic sequence to obtain directionality data;

[0084] S23, using the bedtools tool to integrate the narrowpeak files of the five transcription factors with the Hi-C loop data to generate a file containing transcription factor information and overlapping regions;

[0085] S24. Sort and process the data to finally construct the arrangement of key transcription factors within the chromatin loop.

[0086] In step S3, when building the classification model, 80% of the samples are randomly selected as training sets and 20% as independent test sets. The pre-trained model of BERT-Base-Uncased is used for fine-tuning and binary classification.

[0087] Among them, BERT is a natural language processing model, and BERT-Base-Uncased is a specific variant of the BERT model.

[0088] BERT, or Bidirectional Encoder Representations from Transformers, is a natural language processing model developed by Google that aims to provide powerful pre-trained representations for a wide range of language understanding tasks.

[0089] In this experiment, the input to the model is the sequence of key transcription factors within chromatin loops.

[0090] In detail, the model building includes the following steps:

[0091] S31. First, define a preprocessing function to tokenize the sequence and fill it with a fixed length.

[0092] S32. Then define the training parameters and evaluation indicators.

[0093] Through the above steps, a classification model of CTCF chromatin loops based on BERT-Base-Uncased was established, and the classification model was used to classify the key transcription factor sequences in chromatin loops.

[0094] Step S4 specifically includes the following steps:

[0095] S41. Obtain the predicted probabilities and true labels of the LC model and BERT-Base-Uncased models to calculate the AUROC and AUPRC values.

[0096] S42, calculating logistic regression of TPR and FPR based on different thresholds;

[0097] S43. Use FPR as the x-axis and TPR as the y-axis to draw the ROC curve;

[0098] S44. Measure the performance of the classifier and determine the accuracy of the classification results.

[0099] like Figure 3 As shown in the figure, the left figure is the ROC curve, with the horizontal axis being the false positive rate (FalsePositive Rate) and the vertical axis being the true positive rate (TruePositive Rate). The blue curve in the figure represents the LC model, and the orange curve represents the Bert-base-uncased model. A higher AUROC indicates that the classification performance of the Bert-base-uncased model is better than that of the LC model. The right figure is the PRC curve, with the horizontal axis being the recall rate (Recal l) and the vertical axis being the precision rate (Precision). A higher AUPRC also indicates that the Bert-base-uncased model performs better when dealing with imbalanced data sets.

[0100] The area under the ROC curve is called AUROC, which can measure the performance of the classifier and determine the accuracy of the classification results. Since the data set of this experiment is unbalanced, the ratio of positive and negative sets is 1:65, and the positive samples are far less than the negative samples, the PRC curve is used to evaluate the model performance. The horizontal axis of the PRC curve is the recall rate, and the vertical axis is the precision.

[0101] In step S41, the formula for calculating AUROC and AUPRC values ​​is:

[0102]

[0103] Among them, TP, FP, TN and FN represent the number of true positives, false positives, true negatives and false negatives respectively. TPR is the true positive rate, that is, the proportion of positive samples correctly identified as positive to all positive samples; FPR is the false positive rate, that is, the proportion of negative samples incorrectly identified as positive to all negative samples.

[0104] In summary, the principle of this embodiment is: by collecting five transcription factors in the SC-βorg stage of pancreatic cell differentiation, namely: CTCF, NIBPL, WAPL, YY1, and RAD21, among which CTCF has directionality, recorded as CTCF+ and CTCF- ChIP-seq data to describe the distribution of different transcription factors in the genome, each transcription factor is represented by a different letter, and the transcription factor sequence is formed according to its binding position encoding, and Hi-C experimental data is used to mark whether any two CTCF binding sites form a chromatin loop, and the CTCF chromatin loop is predicted using the Bert pre-training model.

[0105] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A method for predicting CTCF chromatin loops based on transcription factor sequences in a loop extrusion model, characterized in that: The method comprises the following steps: S1. Build a dataset. S2, key transcription factor sequence conversion; S3. Establish a CTCF chromatin loop classification model based on key transcription factor sequences; S4. Model evaluation.

2. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 1, characterized in that: In step S1, chromatin loops are detected using Hi-C experimental data, and the position coordinates of transcription factors in the genome are obtained through ChIP-seq data of transcription factors in the SC-βorg stage of pancreatic cell differentiation.

3. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 1, characterized in that: Step S1 includes the following steps: S11, obtain all CTCF combination patterns within 1 Mb; S12, select the chromatin loops verified by Hi-C experiments in all CTCF pairs as the positive set; S13, remove the positive set Positive Set and Bin Set trash data, and the remaining is the negative set negative_pool.

4. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 3, characterized in that: Step S11 includes the following steps: S111, firstly, all CTCFs in the SC-βorg stage of pancreatic cell differentiation were combined in pairs within each chromosome; Step S12 includes the following steps: S121, remove all CTCF pairs outside the 1Mb range, retain the starting coordinates of the latter CTCF - the ending coordinates of the former CTCF, and call this data set 1mb_ctcf_pairs, then perform overlap analysis on all 1mb_ctcf_pairs and Hi-C data to find out the regions that are included in 1mb_ctcf_pairs and overlap in Hi-C data; S122, select the CTCF binding site with the highest peak in the overlapping area, which is defined as the positive sample, i.e. PositiveSet; Step S13 includes the following steps: S131, the remaining overlapping area data is put into the Bin Set trash database; S132, remove the positive set Positive Set and Bin Set trash bin data from all 1mb_ctcf_pairs, and the remaining is the negative set pool negative_pool.

5. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 4, characterized in that: Step S2 specifically includes the following steps: S21. Download the CTCF motif from the JASPAR database: MA0139.1.transfac; S22, combine the FASTA file of the CTCF genomic sequence to obtain directionality data; S23, Integrate the narrowpeak files of the five transcription factors with the Hi-C loop data using the bedtools tool to generate a file containing transcription factor information and overlapping regions; S24. The data were sorted and processed, and finally the arrangement of key transcription factors in the chromatin loop was constructed.

6. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 1, characterized in that: In step S3, when building the classification model, 80% of the samples are randomly selected as training sets and 20% as independent test sets. The pre-trained model of BERT-Base-Uncased is used for fine-tuning and binary classification.

7. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 6, characterized in that: The BERT is a natural language processing model, and the BERT-Base-Uncased is a specific variant of the BERT model.

8. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 7, characterized in that: The model building includes the following steps: S31. First, define a preprocessing function to tokenize the sequence and fill it with a fixed length. S32. Then define the training parameters and evaluation indicators. Through the above steps, a classification model of CTCF chromatin loops based on BERT-Base-Uncased was established, and the classification model was used to classify the key transcription factor sequences in chromatin loops.

9. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 1, characterized in that: Step S4 specifically includes the following steps: S41. Obtain the predicted probabilities and true labels of the LC model and BERT-Base-Uncased models to calculate the AUROC and AUPRC values. S42, calculating logistic regression of TPR and FPR based on different thresholds; S43. Use FPR as the x-axis and TPR as the y-axis to draw the ROC curve; S44. Measure the performance of the classifier and determine the accuracy of the classification results.

10. The method for predicting CTCF chromatin loops based on transcription factor sequences in the loop extrusion model according to claim 9, characterized in that: In step S41, the formula for calculating AUROC and AUPRC values ​​is: Among them, TP, FP, TN and FN represent the number of true positives, false positives, true negatives and false negatives respectively. TPR is the true positive rate, that is, the proportion of positive samples correctly identified as positive to all positive samples; FPR is the false positive rate, that is, the proportion of negative samples incorrectly identified as positive to all negative samples.

Images