Mastervar - Complete Genomics Data Format To Vcf

Question

[Peixe]

Hi,

I have been recently dealing with the Personal Genomes Project, and trying to work with the data. I downloaded the raw data for an individual's whole genome.

The main concern is the format of the data. Complete Genomics frees the genomes of the individuals in its own format; a format called masterVar which looks like this:

#ASSEMBLY_ID    GS000014558-ASM
#COSMIC    COSMIC v48
#DBSNP_BUILD    dbSNP build 132
#GENOME_REFERENCE    NCBI build 37
#SAMPLE    GS01669-DNA_D02
#GENERATED_BY    cgatools
#GENERATED_AT    2012-Sep-28 19:43:38.251270
#SOFTWARE_VERSION    2.0.4.14
#FORMAT_VERSION    2.0
#GENERATED_BY    dbsnptool
#TYPE    VAR-ANNOTATION
>locus    ploidy    allele    chromosome    begin    end    varType    reference    alleleSeq    varScoreVAF    varScoreEAF    varQuality    hapLink    xRef
17    2    all    chr1    11365    11370    ref    =    =                    
302    2    1    chr1    21579    21580    snp    C    T    123    123    VQHIGH        dbsnp.83:rs526642
302    2    2    chr1    21579    21580    snp    C    T    153    153    VQHIGH        dbsnp.83:rs526642

They provide some tools to work on it and I tried to convert to vcf with this tool, but what I get is some kind of funny vcf, with duplicated entries and inconsistent information.

Has anyone dealt with it before?

Thanks in advance!

P.

[galina.erikson]

Hi, I'm dealing with the same issue. Did you figure out any way to convert Complete Genomics to vcf or to plink ped format without bugs?

Thank you.

[Peixe]

Hi, Unfortunately I was not able to make it work and quited by now. As said, I tried different conversion tools, but all returned a very weird file with clear errors compared to original. I am really surprised that no further information on this issue explaining the thing a bit more could be found... Anyway, if you get to know anything else, let me know about it, Best,

Answer 1

[ash.avalon]

You could use the CGAtools and use the following command, I left out the CGA_CEHQ and CGA_CEGL field

/software/cgatools/current/bin/cgatools mkvcf --beta --reference build36.crr \
            --source-names masterVar \
            --genome-root $root \
            --master-var <(bzcat $masterVar) \
            --field-names GT,PS,NS,AN,AC,AF,SS,FT,CGA_XR,CGA_ALTCALLS,CGA_FI,GQ,HQ,EHQ,GL,DP,AD,CGA_RDP,CGA_ODP,CGA_OAD,CGA_ORDP,CGA_PFAM,CGA_MIRB,CGA_RPT,CGA_SDO,CGA_SOMC,CGA_SOMR,CGA_SOMS,CGA_SOMF,GT,CGA_GP,CGA_NP,CGA_CP,CGA_PS,CGA_CT,CGA_TS,CGA_CL,CGA_LS,CGA_LAFS,CGA_LLAFS,CGA_ULAFS,CGA_SCL,CGA_SLS,CGA_LAFP,CGA_LLAFP,CGA_ULAFP,GT,FT,CGA_IS,CGA_IDC,CGA_IDCL,CGA_IDCR,CGA_RDC,CGA_NBET,CGA_ETS,CGA_KES,GT,FT,CGA_BF,CGA_MEDEL,MATEID,SVTYPE,CGA_BNDG,CGA_BNDGO,CGA_BNDMPC,CGA_BNDPOS,CGA_BNDDEF,CGA_BNDP

[MAPK]

Where do I find the full details of all these field descriptions?

[kenny.bryan]

This pretty much worked for me - however I had to first convert my genome.fa reference into the CRR format using another cgatool:

$cgatools fasta2crr --input build36.fa.bz2 --output build36.crr

see docs http://cgatools.sourceforge.net/docs/1.4.0/cgatools-install.pdf

Also for those not familiar with linux you need to assign your filename to masterVar variable (masterVar=my_file.tzv.bz2) for the <(bzcat $masterVar) part to work ( Note: not to be confused with the masterVar string in --source-names masterVar that tells the program the conversion type)

Also --genome-root $root was complaining so I left that parameter out (which seems to be OK if you also leave out the CGA_CEHQ and CGA_CEGL as suggested above).

In the process of converting 380 file now and seems to be producing sensible VCF files

(For the record the masterVar2VCF_rev1_ShanYang.zip converter on the complete genomics website did not work for me either. producing corrupt files)