Gatk3.2.2 Summary

After several days of exploration and online data query, the gatk software is a little cautious. The following is a summary:

1. It is best to use the FASTA file to locate the data on the chromosome. You do not need to comment out the VCF file (GVF). However, if you use the VCF file, ensure the following conditions:

1) The number of VCF chromosome and FASTA chromosome must be consistent in sequence.

2) VCF sites must be sorted in ascending order.

3) The VCF base may have other symbols, such as "~". And so on.

2. Before doing this, use BWA index, createsequencedictionary. jar in Picard, and faidx in samtools to index the FASTA file. It is best to create an index in the same folder of FASTA.

3. When BWA is compared, it is best to add the-R parameter: "@ RG \ TID: Name \ TLB: Name \ TPL: illumina \ TSM: Name". In order not to add header files in the future

4. reordersam. jar in Picard is used to correct that the header file of your Sam file is consistent with that of FASTA. If it is consistent, you do not need to do this step.

5. if Picard is used to process any program of the binary red Sam or BAM of BWA, it is best to add validation_stringency = lenient, because when there is a pair of red reads to the end of the chromosome, if another Picard cannot be identified, an error will be reported to terminate the operation.

6. If the call variant sample is merged, gatk has two threads. nt indicates that several samples use one CPU. NCR indicates that one sample uses several CPUs.

7. The reducereads program is no longer supported after gatk 3.0.

Quality value correction and variant correction are not performed at this stage. First, the data volume is required to be large. If the data size is smaller than m, reads should not be performed. Second, it is difficult for current commercial projects to handle such problems, in addition to human projects, there are a lot of corresponding comments.

Of course there are other methods for correction. For example, if they are consistent with the results of samtools mpileup, they are considered reliable.