We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

Genotyping a second set of sites

jfarrelljfarrell Member ✭✭

What is the proper way to genotype an multiple alternative set of sites (eg GONL sites and 1000 genomes)? I am getting a vcf file that is not quite right.

  1. The pipeline is first run (pre, discovery, genotyping).
  2. Then I run an alternative set of sites with the GONL sites and then the 1000 genomes sites. I use the same meta and output directory.
  3. For the 2nd set of sites based on the 1000 genomes sites, I am getting a vcf that has more than the specified sites (sites from previous genotyping is found in the vcf). See below chromosomes 1-22,X is followed by 9-22,X,Y. These later sites match the counts from the GONL sites. It seems to be appending sites from the previous run.

      count #CHROM
    

    1117 1
    1141 2
    941 3
    913 4
    830 5
    850 6
    852 7
    732 8
    644 9
    651 10
    724 11
    683 12
    479 13
    467 14
    417 15
    466 16
    448 17
    395 18
    463 19
    342 20
    165 21
    251 22
    449 X
    905 9
    1331 10
    1281 11
    1371 12
    972 13
    872 14
    717 15
    874 16
    906 17
    784 18
    809 19
    615 20
    446 21
    458 22
    1022 X
    33 Y

Best Answer

Answers

  • jfarrelljfarrell Member ✭✭

    I was expecting the -startFromScratch to address any temp files but that did not seem to help. So it sounds like I should point to the original meta data directory, but have the output go to a new run directory when genotyping a new set of sites.

  • bhandsakerbhandsaker Member, Broadie ✭✭✭✭

    Yes, that's exactly right. -startFromScratch is a Queue argument that simply means to assume the outputs of the current pipeline need to be recomputed. It won't touch the output of other pipelines that may have been run in the same run directory. You always need to keep the run directories separate.

Sign In or Register to comment.