VarScan.v2.4.6.description.txt

RELEASE NOTES FOR VARSCAN V2.4.6

28-Mar-2023

VarScan v2.4.6 is the current release
VarScan v2.4.0 was the first release to VarScan's new home at GitHub, http://dkoboldt.github.io/varscan/
VarScan v2.3.9 and prior releases will persist on SourceForge, as will the support forums and other resources. 

LICENSE
VarScan 2 is free for non-commercial use by academic, government, and non-profit/not-for-profit institutions.
A commercial version of the software is available, and licensed through the Office of Technology Management at
Washington University School of Medicine. For more information please visit their website at: 
https://otm.wustl.edu/for-industry/tools/
or contact them by e-mail: 
otm@wustl.edu

VERSION 2.4.6 CHANGES
This release includes important fixes/improvements to the mpileup2somatic multi-tumor-sample calling functionality introduced in v2.4.5.
-The VCF output format has changed, specifically the INFO, FORMAT, and genotype fields

VarScan processSomatic functionality is compatible with mpileup2somatic VCF output, but provides only basic functionality. 
-It will properly output variants according to overall somatic status

It also contains a bug fix for the parsing of very long (>9999 bp) indels from mpileup input.

NOTE: mpileup2somatic is a new feature, so please send feedback if you use it: daniel.koboldt (at) nationwidechildrens [dot] org.


REMINDER: PLEASE USE THE FALSE POSITIVE FILTER (fpfilter)
We recommend processing initial VarScan output with processSomatic, and then applying fpfilter to high-confidence Somatic calls.
Although the "somaticFilter" function will continue to be available, we recommend using the above approach instead.

The scientific basis of this filter is described in the VarScan 2 publication. It will improve
the precision of variant and mutation calling by removing artifacts associated with short-read alignment.
-For somatic mutations, generate bam-readcounts with the Tumor BAM. For LOH and Germline, generate readcounts with the Normal BAM
-For de novo mutations (trio calling), generate readcounts with the child BAM.
The filter requires the bam-readcount utility: https://github.com/genome/bam-readcount

USAGE: java -jar VarScan.jar fpfilter [variant file] [readcount file] OPTIONS
	variant file - A file of SNPs or indels in VarScan-native or VCF format
	readcount file - The output file from bam-readcount for those positions
	***For detailed filtering instructions, please visit http://varscan.sourceforge.net***

	OPTIONS:
	--output-file		Optional output file for filter-pass variants
	--filtered-file		Optional output file for filter-fail variants
	--dream3-settings	If set to 1, optimizes filter parameters based on TCGA-ICGC DREAM-3 SNV Challenge results
	--keep-failures		If set to 1, includes failures in the output file

	FILTERING PARAMETERS:
	--min-var-count		Minimum number of variant-supporting reads [4]
	--min-var-count-lc	Minimum number of variant-supporting reads when depth below somaticPdepth [2]
	--min-var-freq		Minimum variant allele frequency [0.05]
	--max-somatic-p		Maximum somatic p-value [1.0]
	--max-somatic-p-depth	Depth required to test max somatic p-value [10]
	--min-ref-readpos	Minimum average read position of ref-supporting reads [0.1]
	--min-var-readpos	Minimum average read position of var-supporting reads [0.1]
	--min-ref-dist3		Minimum average distance to effective 3' end (ref) [0.1]
	--min-var-dist3		Minimum average distance to effective 3' end (var) [0.1]
	--min-strandedness	Minimum fraction of variant reads from each strand [0.01]
	--min-strand-reads	Minimum allele depth required to perform the strand tests [5]
	--min-ref-basequal	Minimum average base quality for ref allele [15]
	--min-var-basequal	Minimum average base quality for var allele [15]
	--min-ref-avgrl		Minimum average trimmed read length for ref allele [90]
	--min-var-avgrl		Minimum average trimmed read length for var allele [90]
	--max-rl-diff		Maximum average relative read length difference (ref - var) [0.25]
	--max-ref-mmqs		Maximum mismatch quality sum of reference-supporting reads [100]
	--max-var-mmqs		Maximum mismatch quality sum of variant-supporting reads [100]
	--max-mmqs-diff		Maximum average mismatch quality sum (var - ref) [50]
	--min-ref-mapqual	Minimum average mapping quality for ref allele [15]
	--min-var-mapqual	Minimum average mapping quality for var allele [15]
	--max-mapqual-diff	Maximum average mapping quality (ref - var) [50]


DREAM-3 SETTINGS FOR FPFILTER
Please note the --dream3-settings parameter for fpfilter, which (if set to 1) will optimize the 
false positive filter settings based on the fine-tuning we did for the TCGA-ICGC DREAM-3 
SNV Challenge. See the "in silico 3" dataset described here:
https://www.synapse.org/#!Synapse:syn312572/wiki/62018
This dataset modeled 100% tumor purity, but three subclones at 50%, 33%, and 20% variant allele frequency.
Optimal VarScan settings were established as follows:
	For SAMtools:
		mpileup -B (disables BAQ)

	For VarScan somatic:
		--min-coverage 3 --min-var-freq 0.08 --p-value 0.10 --somatic-p-value 0.05 --strand-filter 0

	For VarScan fpfilter:
		--min-var-count = 3
		--min-var-count-lc = 1
		--min-strandedness = 0
		--min-var-basequal = 30
		--min-ref-readpos = 0.20
		--min-ref-dist3 = 0.20
		--min-var-readpos = 0.15
		--min-var-dist3 = 0.15
		--max-rl-diff = 0.05
		--max-mapqual-diff = 10
		--min-ref-mapqual = 20
		--min-var-mapqual = 30
		--max-var-mmqs = 100
		--max-ref-mmqs = 50

CITING VARSCAN
If you use VarScan, please note the version number and cite this publication along with the 
version-appropriate URL:

Koboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, Miller CA, Mardis ER, Ding L, Wilson RK. 
VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. 
Genome Res. 2012 Mar;22(3):568-76. doi: 10.1101/gr.129684.111. 

https://github.com/dkoboldt/varscan (v2.4.0 and beyond)
or
http://varscan.sourceforge.net (v2.3.9 and before)