Data filtering

To remove the noise from the sequencing data, we first filtered the data. The original data filtering method is as follows：

1. Filter out the adaptor sequence of reads;

2. When the low-quality base number in the single-end sequencing reads exceeds 50% of the reads, this pair of reads will be removed.

3. When the content of N base in the single-end sequencing reads exceeds 10% of the reads, this pair of reads will be removed.

After filtration, "cleandata" was obtained and the sequencing data were statistically analyzed, including the number of sequencing reads, data yield, and distribution of quality value, etc.

​

Data aligning
All cleanreads were aligned to the human reference genome (GRCh37/HG19) using the Burrows-Wheeler Aligner alignment software (BWA V0.7.12). The sequencing data of each lane were compared, and the read group ID was added in the comparison result.

​

Base Quality Score Recalibration (BQSR)
Variation detection methods rely on the quality values of base sequencing. Systematic errors caused by various sequencing instruments may cause the base quality score to be too high or too low. Therefore, it is necessary to correct the base quality score to obtain a more accurate base quality score, so as to improve the accuracy of mutation detection.

 

Filtration variation result
When obtaining original variation sets including SNP and InDel, it is necessary to obtain high-quality and highly reliable variation sets for downstream analysis by filtering. Here, the VQSR method based on a machine learning algorithm is adopted to filter the original variation set. GATK VQSR uses the known and high-quality variation set as the training set and the real set and establishes a prediction model to filter out the false variation. The output VCF result is marked as"PASS" of the SNPs and InDel, a reliable set of mutations that PASS the filter.

​

Copy number variation detection
CNVnator\[8\] v0.2.7 based on the depth signal algorithm was used to detect copy number variation. In this method, the genome was segmented into non-overlapping Windows of equal length, and the number of reads aligned on each window was standardized as a depth signal to detect copy number variation.

​

Structural variation detection
The procedure detected 5 types of structural variation (SV), (1) interchromosomal translocation (CTX), (2) interchromosomal translocation (ITX), (3) INV, (4) deletion (DEL), and (5) insertion (INS).

 

Annotation and prediction of variation results
After obtaining a highly reliable, high-quality variation set, use SnpEff to perform annotation：

gene-based annotation：\identify SNPs/InDels resulting in an alternation of protein-coding and amino acid sequences.

database-based annotations：Identify mutations that appear in the dbSNP library (v141), or whose minor allele frequency (MAF) in the 1000 genome project is less than 1%, or whose coding region is not synonymous with the mutation 

Whether the SIFT value of SNPs is less than 0.05, or whether the conservative predictive value of variation GERP++ value is greater than 2, or other annotation information.