CN104131093B - DNase high-throughput sequencing detection signal processing method of DNA protein binding site - Google Patents

Abstract

The invention discloses a DNase high-pass sequencing detection signal preprocessing method of a DNA protein binding site. Including the following steps: Obtain basic gene information, DNase-Seq high-pass sequencing detection data and ChIP-Seq high-pass sequencing monitoring data of DNA protein binding sites; evaluate the quality of DNase-Seq high-pass sequencing detection data, and screen out reliable sequencing Data; keep only the sequencing start position that directly reflects the protein binding site for each credible sequencing data; obtain the DNase-Seq detection sample data set; normalize the DNase-Seq detection sample data set; DNase-Seq The test sample data set is subdivided; the data in the two subsets are summed vertically from the front and back respectively to complete the operation. The invention greatly improves the recognition accuracy and recognition resolution of the DNA protein binding site.

Description

DNase high-throughput sequencing detection signal processing method of DNA protein binding site

technical field

The invention belongs to a detection signal processing method of a DNA protein binding site, in particular to a DNase high-pass sequencing detection signal processing method of a DNA protein binding site.

Background technique

At present, the detection of DNA protein binding sites mainly adopts the technique of chromatin immunoprecipitation (Chromatin Immunoprecipitation, ChIP). The ChIP-Seq technology, which combines the results of ChIP experiments with high-throughput sequencing technology, can efficiently detect DNA segments that bind to specific functional proteins on a genome-wide scale. The principle of ChIP-Seq is: firstly, the DNA fragments bound to the target protein are enriched by using the enzyme that specifically binds to the target protein through chromatin immunoprecipitation (ChIP), and then purified and library constructed. Then perform high-throughput sequencing on the enriched DNA fragments, and then accurately map the millions of read sequences obtained from the sequencing to the genome, so as to obtain the DNA segment information that binds the target protein in the whole genome, and then through each This analysis algorithm obtains accurate DNA binding sites of the target protein. However, although the DNA protein binding site analysis method for ChIP-Seq detection data is very mature, this technology also has shortcomings. It cannot be detected without a suitable specific binding enzyme; secondly, only one protein can be detected in one experiment, which is time-consuming, labor-intensive, and expensive, and cannot be used on a large scale; The DNA fragment bound by the target protein is long, and only part of its two ends can be sequenced during sequencing. Since the sequencing region is not the binding site itself, although the resolution of the sequencing data can reach a single base, the binding site of the target protein The highest point positioning resolution can only reach tens of bases.

In response to the above problems, a new DNA protein binding site detection technology has emerged in recent years--DNA protein binding site detection technology based on DNase high-throughput sequencing information, namely DNase-Seq technology. This technique is also called DNase footprinting (DNasefootprinting), which can accurately identify the binding sites of DNA binding proteins on DNA molecules. The principle is: firstly, the DNA is digested with DNase nuclease. Then the DNA region without DNA protein binding will be randomly and evenly cut by DNase nucleoclease, while the DNA region with DNA protein binding will not be specifically cut due to the hindrance of the binding protein. Subsequently, the digested DNA fragments were purified and library constructed, and then sequenced, so as to obtain the enzyme digestion information of DNase nuclease in the whole genome. In the enzyme cleavage information, the enzyme cleavage information at the protein binding site will weaken the specificity, just like leaving footprints on the DNA, so that the binding site of the DNA binding protein on the DNA molecule can be accurately identified. Compared with ChIP-Seq technology, the advantages of the newly proposed DNase-Seq high-throughput sequencing technology are very prominent. First of all, due to its lack of specificity, DNase-Seq can detect the binding sites of all DNA proteins in the whole genome at one time; secondly, due to the one-time detection of all DNA protein binding sites, DNase-Seq technology has greatly It improves the detection efficiency and reduces the detection cost, and makes it possible to detect DNA protein binding sites on a large scale; thirdly, and more importantly, since the starting position of DNase-Seq sequencing is the enzyme cutting position, therefore, DNase-Seq The detection resolution of DNA protein binding sites can reach a single base, such a high resolution is very helpful for follow-up research. Therefore, it is necessary to process the DNase-Seq signal and conduct in-depth research and analysis.

Since the technology of DNase to accurately detect DNA protein binding sites was proposed, DNase technology has been used to provide accurate experimental verification for research results related to DNA binding sites. The processing of its experimental data is also mostly analyzed in the form of simple counting and expressed in the form of intuitive graphics. Until 2006, Crawford and Sabo proposed the DNase-Chip technology in Naturemethods at the same time, using microarray to perform high-throughput measurement of DNase detection signals, thus opening the application of DNase technology for large-scale detection and analysis of protein binding sites within the genome stage. In 2010, Crawford further proposed the DNase-Seq technology that can detect protein binding sites on a genome-wide scale.

After the DNase-Seq technology was proposed, many analysis methods have been produced one after another. In 2010, Chen used DNase-Seq data to analyze DNA protein binding sites based on dynamic Bayesian network. In 2011, Fletez analyzed DNA protein binding sites using DNase-Seq data based on support vector machines. In 2012, Pique proposed the CENTIPEDE method to analyze the statistical characteristics of DNase-Seq data at the DNA protein binding site by designing a rigorous statistical model, and then identify the DNA protein binding site. In 2013, Jason proposed the Wellington method to identify DNA protein binding sites using the DNase-Seq data characteristics when the DNA protein binding site chromosome region is open. In 2014, Sherwood used the unique amplitude and shape characteristics of the DNA protein binding site region to identify DNA protein binding sites.

However, various analysis methods currently proposed use the preprocessing method of ChIP-Seq detection data to preprocess the DNase-Seq detection data. But in essence, due to the significant difference in the detection principle, the resolution of DNase-Seq data for the detection of DNA protein binding sites is much higher than that of ChIP-Seq data.

Contents of the invention

The purpose of the present invention is to provide a DNase high-throughput sequencing detection signal processing method for DNA protein binding sites that can provide recognition accuracy and recognition resolution of DNA protein binding sites.

The present invention is achieved through the following technical solutions:

The DNase high-throughput sequencing detection signal preprocessing method of the DNA protein binding site comprises the following steps:

Step 1: Obtain the basic information of the gene, including the base sequence of the DNA genome and the position information of the gene on the DNA, obtain the DNase-Seq high-throughput sequencing detection data of the DNA protein binding site and the high-throughput ChIP-Seq Sequencing detection data;

Step 2: Assess the quality of DNase-Seq high-pass sequencing detection data, screen out credible sequencing data with a quality score of more than 20 base sites, and find the source of each credible prediction data in the genome through mapping;

Step 3: Keep only the sequencing start position that directly reflects the protein binding site for each credible sequencing data, and obtain the updated DNase-Seq data;

Step 4: Calculate the number of updated DNase-Seq data points at each DNA base site as the DNase-Seq detection value of the DNA base site; use ChIP-Seq data to obtain the presence of related DNA proteins The region of the binding site, extract the complete DNase-Seq detection value in the region, and obtain the DNase-Seq detection sample data set;

Step 5: Normalize the DNase-Seq detection sample data set, that is, divide the DNase-Seq detection value of each DNA base site by the DNase-Seq detection value of all DNA base sites in the DNase-Seq detection sample data set. The sum of Seq detection values;

Step 6: subdivide the DNase-Seq detection sample data set;

The DNase-Seq detection sample data set is divided into positive-strand positive sequencing subset, positive-strand negative sequencing subset, negative-strand positive-sequencing subset and negative-strand negative sequencing subset. The subsets are merged into the positive detection data subset of the DNA protein binding site through correlation alignment, and the positive strand negative sequencing subset and the negative strand positive sequencing subset are merged into the back detection data of the DNA protein binding site through correlation alignment subset of data;

Step 7: Vertically sum the data in the front detection data subset and the rear detection data subset respectively from the front and rear directions, and complete the operation.

Beneficial effects of the present invention:

Due to the significant difference in detection principles, the resolution of DNase-Seq data for the detection of DNA protein binding sites is much higher than that of ChIP-Seq data. Identification accuracy and identification resolution of protein binding sites.

At the same time, through the processing of DNase-Seq detection data, the detection information of DNA protein binding sites can be highlighted, laying the foundation for the subsequent high-precision identification of protein binding sites.

Description of drawings

Fig. 1 Block diagram of DNase high-throughput sequencing detection signal preprocessing method of DNA protein binding site;

The operation process of Fig. 2 conventional treatment;

The operation process of Fig. 3 special treatment;

Figure 4 DNase-Seq positive strand positive sequencing detection information of DNA protein binding sites, ATF1 protein in K562 cells;

Figure 5 DNase-Seq negative strand negative sequencing detection information of DNA protein binding sites, ATF1 protein in K562 cells;

Figure 6 DNase-Seq positive-strand negative sequencing detection information of DNA protein binding sites, ATF1 protein in K562 cells;

Figure 7 DNase-Seq negative strand positive sequencing detection information of DNA protein binding sites, ATF1 protein in K562 cells;

Figure 8 DNase-Seq positive detection information of DNA protein binding sites, ATF1 protein in K562 cells;

Figure 9 DNase-Seq back detection information of DNA protein binding sites, ATF1 protein in K562 cells.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Obtain the DNase high-throughput sequencing detection data of DNA protein binding sites, conduct in-depth analysis and targeted processing, and finally achieve the purpose of highlighting the detection information of protein binding sites. As shown in Figure 1, it specifically includes the following steps:

1. Data Acquisition

We obtain the basic information of genome base sequence from the international biological information website UCSC; obtain the basic information of DNA protein from the TRANSFAC database of BIOBASE company. The DNase-Seq and ChIP-Seq detection data of DNA protein binding sites are all from the data generated by the ENCODE project published on the UCSC website. Among them, the DNase-Seq detection data was downloaded from the data provided by the DUKE laboratory and the UW laboratory under hg19 conditions on the ENCODE website; the ChIP-Seq detection data was downloaded from the data provided by the SYDH laboratory.

2. Conventional pretreatment part

As shown in Figure 2, the DNase-Seq high-throughput sequencing detection data of DNA protein binding sites are generally detected and generated using the high-throughput sequencing platform of Illumina because of the short-sequence sequencing method. The data generated by the sequencing platform complies with the FASTQ format, that is, each read (read) of the sequencing data is stored in four rows. Among them, the first line is the relevant information of the sequencing platform, starting with ""; the second line is the sequence information; the third line starts with the "+" sign, and the latter is the same as the first line, sometimes omitted; the fourth line is the base detection quality score corresponding to the sequencing sequence. Wherein, each base site has a corresponding quality score. Here, it is required that the quality score of all base positions in the credible sequencing data should be above 20, which means that the error probability of any base position is below 1%.

Then comes the mapping of sequencing data, that is, finding the source of each piece of sequencing data in the genome through mapping. Because the sequencing data of the Illumina sequencing platform have the same length, it is suitable to use the BWA alignment software to realize the mapping from the sequencing data to the genome.

After sequencing data mapping, it is necessary to analyze the mapping quality of each piece of data. It is required that the mapping quality MAPQ score of reliable sequencing data should be above 20 (which means that the probability of sequencing data mapping error is below 1%), and the number of mismatched bases should be below 2.

3. Special pretreatment part

As shown in Figure 3, first of all, since the starting position of DNase-Seq sequencing is the restriction position, and can reflect the DNA protein binding site at single-base resolution, in the preprocessing stage of DNase-Seq detection data, Each piece of DNase-Seq detection data can be changed into a point, that is, only the sequencing start point that directly reflects the protein binding site is retained, thereby effectively highlighting the binding site information of the DNA protein.

Secondly, after preprocessing the DNase-Seq data into points, calculate the number of DNase-Seq data points at each DNA base site as the DNase-Seq detection value at the base site . The ChIP-Seq data was then used to obtain the regions where the binding sites of related DNA proteins existed. If it exists (the resolution is only a few tens of bases), then extract the complete DNase-Seq detection value of this region, and accurately determine the DNA protein binding site through the base propensity of the related protein, and the resolution should reach a single base. In this way, a set of DNase-Seq detection sample data sets for which the DNA protein binding site must exist can be obtained.

In addition, in order to avoid inconsistencies in the contribution of different samples in the subsequent analysis process, the sample data should also be normalized and preprocessed. That is, the DNase-Seq detection value of each base site in the sample should be divided by the sum of the DNase-Seq detection values of all base sites in the sample area.

Subsequently, the sample set is subdivided. Firstly, different samples were divided into DNA positive strand and negative strand according to the location of the DNA protein binding site. Second, each type of sample is divided into DNA positive strand and negative strand according to the orientation of its sequencing reads. In this way, all samples in the sample set are divided into four parts: positive strand positive sequencing, positive strand negative sequencing, negative strand positive sequencing, and negative strand negative sequencing. Among them, positive strand positive sequencing means that when the DNA protein binding site is on the positive strand, the detection value of the DNA protein binding site is detected from the positive strand; positive strand negative sequencing means that the DNA protein binding site is on the positive strand When the DNA protein binding site is detected from the negative strand, the meaning of negative strand positive sequencing is that when the DNA protein binding site is on the negative strand, the detection value of the DNA protein binding site is detected from the positive strand . Negative strand negative sequencing means that when the DNA protein binding site is on the negative strand, the detection value of the DNA protein binding site is detected from the negative strand. According to the DNase-Seq biological detection mechanism of the DNA protein binding site, the positive-strand positive sequencing and the negative-strand negative sequencing both detect the DNA protein binding site from the front, while the positive-strand negative sequencing and negative-strand positive sequencing both detect the DNA protein binding site from the back The DNA protein binding site is detected, therefore, the positive strand positive sequencing and the negative strand negative sequencing can be combined into a positive detection data subset of the DNA protein binding site through correlation alignment, and the positive strand negative sequencing and the negative strand positive sequencing can be Merged into a subset of backside assay data for DNA protein binding sites by relative alignment.

Finally, all samples were longitudinally summed from the front and back directions, so as to highlight the overall detection information of DNase-Seq on the DNA protein binding site on the basis of statistical noise removal. This information can be used to extract and form the unique DNase-Seq signal pattern of related DNA proteins, and be used for genome-wide identification and detection of the DNA protein binding sites with high precision and high resolution in subsequent experiments.

4. Experimental verification

The DNase-Seq and ChIP-Seq detection data of the DNA-binding site of the DNA-binding protein ATF1 in the K562 cell line were obtained from the internationally published UCSC biological information website. Figures 4 to 9 show the results of the DNase-Seq detection data of ATF1 protein processed by the preprocessing analysis method of the present invention. Among them, the horizontal axis is the base position of the DNA region where the DNA protein binding site is located, the left side is the 5' end of the DNA region, the right side is the 3' end, and the vertical axis is the DNase-Seq detection value. For clear display, as shown in Figure 4 and Figure 5, the negative-strand negative sequencing and positive-strand negative sequencing are horizontally reversed compared with the conventional display mode. The area between the two vertical lines in the figure is the binding site of ATF1 (DNA protein binding is directional, it should be from the upstream 5' end of the DNA to the downstream 3' end).

It can be clearly seen that in Figures 4 to 7, looking from the 5' end to the 3' end of the DNA, between positive-strand positive sequencing and negative-strand negative sequencing, and between positive-strand negative sequencing and negative-strand positive sequencing, The DNase-Seq detection signal at its DNA binding site is basically the same; but between positive strand positive sequencing, negative strand negative sequencing and positive strand negative sequencing, negative strand positive sequencing), the DNase-Seq at its DNA binding site Seq detection signals are different. This shows that through the subdivision of DNase-Seq samples, the detection information of DNA protein binding sites is effectively highlighted, and the purpose of preprocessing is achieved.

Finally, the positive-strand positive sequencing data and the negative-strand negative sequencing data are aligned and added to reflect the detection of the positive binding of the DNA binding site, as shown in Figure 8; correspondingly, the positive-strand negative sequencing data and The positive and negative strand sequencing data were aligned and added to reflect the detection of the back binding of the DNA binding site, as shown in FIG. 9 . After this processing, the detection information of the DNA protein binding site is further highlighted.

The experimental results show that the preprocessing method of the DNase-Seq signal proposed by the present invention effectively highlights the detection information of the DNA protein binding site, which is used to accurately extract the recognition pattern of the DNA protein binding site and further realize the identification of the DNA protein binding site. Identify a good foundation.

Claims (1)

1. The preprocessing method of the DNase high-throughput sequencing detection signal of the DNA protein binding site, is characterized in that, comprises the following steps: Step 1: Obtain the basic information of the gene, including the base sequence of the DNA genome and the position information of the gene on the DNA, obtain the DNase-Seq high-throughput sequencing detection data of the DNA protein binding site and the high-throughput ChIP-Seq Sequencing detection data; Step 2: Assess the quality of DNase-Seq high-throughput sequencing detection data, screen out credible sequencing data with a quality score of more than 20 base sites, and find the source of each credible prediction data in the genome through mapping; Step 3: Keep only the sequencing start position that directly reflects the protein binding site for each credible sequencing data, and obtain the updated DNase-Seq data; Step 4: Calculate the number of updated DNase-Seq data points at each DNA base site as the DNase-Seq detection value of the DNA base site; use ChIP-Seq data to obtain the presence of related DNA proteins The region of the binding site, extract the complete DNase-Seq detection value in the region, and obtain the DNase-Seq detection sample data set; Step 5: Normalize the DNase-Seq detection sample data set, that is, divide the DNase-Seq detection value of each DNA base site by the DNase-Seq detection value of all DNA base sites in the DNase-Seq detection sample data set. The sum of Seq detection values; Step 6: subdivide the DNase-Seq detection sample data set; The DNase-Seq detection sample data set is divided into positive-strand positive sequencing subset, positive-strand negative sequencing subset, negative-strand positive-sequencing subset and negative-strand negative sequencing subset. The subsets are merged into the positive detection data subset of the DNA protein binding site through correlation alignment, and the positive strand negative sequencing subset and the negative strand positive sequencing subset are merged into the back detection data of the DNA protein binding site through correlation alignment subset of data; Step 7: Vertically sum the data in the front detection data subset and the rear detection data subset respectively from the front and rear directions to complete the operation.