These VCF files were called on the full project bam files derived from
the 20100311 sequence.index files, using the QCALL SNP finder.

Only the five populations with the most samples were called, each
independently from the others.  Samples without HapMap3 genotype data
were not called.

The calls are not directly comparable to pilot calls, for multiple reasons:
  - they are only the product of one caller, not 2-or-3 calls from 3 callers
  - they are only filtered for call quality, not for the following filters used for the pilot
    - depth
    - proximity to indels (nor locally realigned to avoid possible new indels)
    - too high a fraction of mapping quality 0 Illumina reads

We suggest these calls may be used to provide additional genotypes at
previously known or called sites (remembering these are also genotypes
from only one caller not consensus genotypes as for the pilot), or to
provide candidate variable sites for other analyses.  They should not
be used as final SNP sites or genotypes for population genetic studies.

Basic statistics about these calls, with comparison to the 0912 QCALL 
calls from pilot data that contributed to the final pilot call set.

        #num calls      VCF Dec-2009     OverLap

CEU     10250970        9535019          9017704

JPT     9047387         7503638 (*)

CHB     7045299

TSI     10611134

YRI     15274485        12062137

(*): JPT+CHB 

MD5 checksums:
3711f27bd5c714c40968e0766f618be7  CHB.March2010.VCF.gz
bd53cc4b656f99c9be519eaaf682d95b  JPT.March2010.VCF.gz
953a95bc9983525417a61e3d9938ed80  TSI.March2010.VCF.gz
b03cb53c2669d63f50f16215ae3e5765  YRI.March2010.VCF.gz
790601958f1e53c1d83c793f136be6f9  CEU.March2010.VCF.gz

Quang Le, Richard Durbin, Sendu Balasubramanian, June 2010