clinical-genomics / balsamic Goto Github PK

• Contest
• Annotation

Describe what needs to done, similar to figure 2 in this document: https://www.accessdata.fda.gov/cdrh_docs/reviews/den170058.pdf

Vardict reports bunch of complex variants that need to be either decomposed or put into another batch of variants.

implementation of CollectInsertSizeMetrics results from Picard in results.

from sbatch set -D, --workdir=directory for job submissions

1. TMB was defined as the number of somatic, coding, base substitution, and indel mutations per megabase of genome examined. https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-017-0424-2
   1.1. Non-coding alterations were not counted.
   1.2 Alterations listed as known somatic alterations in COSMIC and truncations in tumor suppressor genes were not counted
   1.3 Alterations predicted to be germline by the somatic-germline-zygosity algorithm were not counted
   1.4 Alterations that were recurrently predicted to be germline in our cohort of clinical specimens were not counted.
   1.5 Known germline alterations in dbSNP were not counted.
   1.6 To calculate the TMB per megabase, the total number of mutations counted is divided by the size of the coding region of the targeted territory.
   1.7 select the table 1 from paper above as a ref for comparison.

2. Tumor mutation burden (TMB), fraction of copy number–altered genome, and gene alterations were compared among patients with DCB and no durable benefit (NDB). http://ascopubs.org/doi/full/10.1200/JCO.2017.75.3384
   2.1 in addition to above, copy number alterations were also counted.

CollectMultipleMetrics can gather multiple metrics, but it might fail on some metrics due to various reasons (e.g. java memory issue). Prepare a list of metrics that we are interested from this.

Manta in single sample mode should only have the following files in their output:

"tumorSV.vcf.gz", "candidateSV.vcf.gz", "candidateSmallIndels.vcf.gz"

while right now it is:

"diploidSV.vcf.gz", "somaticSV.vcf.gz", "candidateSV.vcf.gz", "candidateSmallIndels.vcf.gz"

• Literature review
• A list aligners
• Variant callers
• QC and error correction

• data.table
• optparser
• stargazer

• balsamic run
• balsamic config sample
• balsamic config report
• balsamic report
• balsamic install

https://github.com/Illumina/Pisces

• Single sample mode workflow on top of #20 and #19
• Single sample mode CLI

An additional rule for picard CollectAlignmentSummaryMetrics

Result report filter config is series of filters, configs, and parameters to summarize analysis results as an effort to present a list of actionable targets and discovered new targets.

This aggregated list will generate 4 list of variants for SNV and INDELs:

• High confidence set of MSK-IMPACT pipeline's exact replica; See FDA approved figure 1.
• Low confidence set of MSK-IMPACT pipeline's exact replica, see FDA approved figure 1.
• High confidence set of variants. identified by at least 1 variant caller, AD>=1, VD>=1, and in MVL.
• Low confidence set of variants. These variants are not in MVL, thus low confidence. So instead, a set variant caller specific filters will be applied according to recent research and clinical findings.

VCF union of all three caller:

• These aggregation should have AD and DP
• Caller type
• PASSed variants only

• A list of metrics to export from pipeline
• A list of metrics from lab side
• Table structure matching with statusDB

BED files need interval preparation and run estimation time. A separate mini-analysis to prepare and split bed files can speed up analysis and reduce runtime.

It seems the AF that mutect2 in gatk3.8 is reporting is not matching the actual AD values. This is due to AD show is "unfiltered" AD. And read depth values are just estimates "before" applying the internal filters. broadinstitute/gatk#3808 and thus DP is not reported by Mutect2 in gatk3.8 (however, it is reported in gatk4 version, where AD values are also not matching with gatk3.8). This could be due to internal filtering process that's happening behind the scene.

A solution to get the values used by Mutect2 to add the actually AD to VCF is to invoke Coverage annotation through --annotation command. Note that DP value reported here will still be unfiltered reads of tumor + normal. For actually DP value, AD values need to be summed.

divide very large bam files into manageable sizes

In accordance with https://github.com/Clinical-Genomics/Goals/issues/51

• Conda environment installation script
• Setup.py
• requirements.txt
• Documentation
• add test

Add up tom 1000X instead of 200X that is today.

After 200X, it should be increased by 50X (e.g. 250, 300, etc) to make it possible to plot the data

• update description for fastq tumor and normal
• check for fastq exists or not

A new tool for plotting structural variants: samplot.py: https://github.com/ryanlayer/samplot

The reports coming out from Picard should be sorted row wised.

report mapping quality in mutect2 using https://software.broadinstitute.org/gatk/documentation/tooldocs/3.8-0/org_broadinstitute_gatk_tools_walkers_annotator_RMSMappingQuality.php

• Coverage vs Genomics coordinate
• Single plot with gene vs coverage for each report config sections (read through MSK MVL list)
• normalized sample coverage

• Vardict
• Strelka
• Manta
• Mutect