Pisces VCF Specifications

The following specification document is valid for both somatic VCF and gVCF formatted files.

SDS ID	Specification
VCF-1	The application shall write a header section at the top of the VCF file with the following header lines. These lines shall have the format “##{key}={value}”. The keys and their descriptions are given below:

Key	Description
fileformat	Version of vcf format, which is “VCFv4.1”.
fileDate	Date in YYYYMMDD format.
Source	Application name and version, e.g. “CallSomaticVariants 1.0.0.0”
CallSomaticVariants_cmdline	Command line call for the program, including all arguments.
Reference	File name for reference genome fasta file.
INFO	Description of INFO fields used in the file. There is one INFO header line for each field in the file.
FILTER	Description of FILTER fields used in the file. There is one FILTER header line for each field in the file.
FORMAT	Description of FORMAT fields used in the file. There is one FORMAT header line for each field in the file.
contig	List of processed chromosomes and their lengths. There is a contig header line for each chromosome. Format is “##contig=<ID={chrName},length={length}”

SDS ID	Specification
VCF-2	The application shall write the following INFO and FORMAT lines to the VCF header, if the associated configuration rule is satisfied. These lines shall have the format: “##{Key}=<ID={FieldName},Number={Number},Type={Type},Description={Description}”.

Key	Field name	Number	Type	Description	Configuration Rule
INFO	DP	1	Integer	Total Depth	None
FORMAT	GT	1	String	Genotype	None
FORMAT	GQ	1	Integer	Genotype Quality	None
FORMAT	AD	.	Integer	Allele Depth	None
FORMAT	DP	.	Integer	Total Depth Used For Variant Calling	None
FORMAT	VF	.	Float	Variant Frequency. One number if 0/0 or 0/1. Two numbers for 1/2	None
FORMAT	NL	1	Integer	Applied BaseCall Noise Level	Debug mode enabled, or outputting bias files, or strand bias threshold < 1
FORMAT	SB	1	Float	StrandBias Score	Debug mode enabled, or outputting bias files, or strand bias threshold < 1
FORMAT	NC	1	Float	Fraction of bases which were uncalled or with basecall quality below the minimum threshold	Report no calls enabled

SDS ID	Specification
VCF-3	The application write the following FILTER lines to the VCF header, if the associated configuration rule is satisfied. FILTER lines shall have format “##{Key}=<ID={FieldName}, Description={Description}”.

Key	FieldName	Description	Configuration Rule
FILTER	q{threshold}, e.g. “q20”	Quality below {thresholdValue}	Minimum variant score configured > 0.
FILTER	LowDP	Low coverage (DP tag), therefore no genotype called	Minimum coverage configured > 0.
FILTER	SB	One of the following, depending on the rule:	Strand bias threshold configured > 0.
FILTER	SB	A)Variant strand bias too high	Strand bias threshold configured > 0.
FILTER	SB	B)Variant support on only one strand	Filter variants on only one strand
FILTER	SB	C)Variant strand bias too high or coverage on only one strand	Three possible rules:

SDS ID	Specification
VCF-4	The application shall write a data section to the VCF file as a tab-delimited table below the header section.
VCF-5	The application shall write a column header line, as below, at the top of the data section of a VCF file. The column header line shall be prefixed by a single “#” and have the following format. The SampleName is set to the input BAM file name (without extension).

#CHROM POS    ID     REF    ALT    QUAL   FILTER INFO   FORMAT {SampleName}

SDS ID	Specification
VCF-6	By default VCF mode, after the column header line, the data section of a VCF file shall have one line per variant allele.
VCF-7	If gVCF mode is selected, after the column header line, the data section of a VCF file shall have one line per allele (reference or variant).
VCF-8	If CrushVcf mode is selected, after the column header line, the data section of a VCF file shall have one line per genomic loci.
VCF-9	For each data line item, the application shall write the following values to the data section of a VCF file, as below:

Column Name	Value
CHROM	Chromosome name
POS	Reference position
ID	Source ID for variant, always “.”. This columns is provided for downstream annotators to update as appropriate.
REF	Reference allele
ALT	Alternate (variant) allele
QUAL	Variant Quality Score
FILTER	“PASS” if no filters. Otherwise, comma-separated list of filter names, e.g. “LowDP,SB”.
INFO	Comma-separated list of INFO name and value pairs, in the format “{name}={value}”. Currently only supporting DP INFO field, e.g. “DP=500”.
FORMAT	Colon-separated list of field names, e.g. “GT:GQ:AD”.
{SampleName}	Colon-separated list of FORMAT field values.

SDS ID	Specification
VCF-10	For each data line time, the application shall write the following FORMAT fields to the data section in a VCF file, if the associated configuration rule is satisfied. As below:

FORMAT Field Name	Field Value	Configuration Rule
GT	Genotype	None
GQ	Genotype Quality Score	None
AD	If variant call, value is “{X},{Y}” where X is the reference depth and Y is the allele depth. If reference call, value is allele depth	None
DP	Total coverage depth used in variant calling	None
VF	Variant frequency	None
NL	Estimated basecall quality	Debug mode enabled, or outputting bias files, or strand bias threshold < 1
SB	Strand bias score	Debug mode enabled, or outputting bias files, or strand bias threshold < 1
NC	No call frequency or fraction	Report no calls enabled