quality check of bam files

Question

[gubrins]

Good afternoon, I'm working with GATK to do the SNP calling of some target capture sequencing data. Right now I'm creating the bam files and I was wondering which are the standard quality measures I should apply to my bam files. I'm aware that I can mark duplicates, but I don't know how this could affect the consequent analyses. Are they removed or they are just marked and I have to do something else? Should I specify some value scores for my bam files?

Thank you very much for your help!

[igor]

Check GATK Best Practices where they address many of your questions.

[gubrins]

Like in here ? I don't fully get GATK BEst Practices

[igor]

They have a lot of different resources. You can also check their events pages, which do not sound like they would be relevant, but they post helpful presentations.

Answer 1

[Pierre Lindenbaum]

. Are they removed or they are just marked and I have to do something else?

they are just marked. Removing PCR duplicates - fastq or BAM?

Each tools in GATK have a default set of Filters for BAM. For example HaplotypeCaller https://gatk.broadinstitute.org/hc/en-us/articles/360037225632-HaplotypeCaller

Read filters

These Read Filters are automatically applied to the data by the Engine before processing by HaplotypeCaller.

NotSecondaryAlignmentReadFilter GoodCigarReadFilter NonZeroReferenceLengthAlignmentReadFilter PassesVendorQualityCheckReadFilter MappedReadFilter MappingQualityAvailableReadFilter NotDuplicateReadFilter => https://gatk.broadinstitute.org/hc/en-us/articles/360037592051-NotDuplicateReadFilter MappingQualityReadFilter WellformedReadFilter

and if needed, any filter can be disabled: https://gatk.broadinstitute.org/hc/en-us/articles/360037225632-HaplotypeCaller#--disable-read-filter

Answer 2

[bruce.moran]

Yes you should quality check your BAMs using CollectHsMetrics to determine on-target reads which is a good indicator of whether the library preparation worked well. Low on-target reads indicates hybridisation failed.

Picard Tools (now part of GATK) has many other tools you can use to QC DNAseq data (e.g. CollectAlignmentSummaryMetrics, CollectMultipleMetrics, CollectSequencingArtifactMetrics, CollectInsertSizeMetrics etc.)

It is a good idea to run fastp or fastqc on fastq files before alignment to determine quality of data initially.

Finally to gather all output into one place MultiQC is a great tool.