Merged bam and unified genotyper

[huma Asif]

Dear All
i have merged my six bam files (one sample run in 6 lanes) using Picard and used GATK pipeline for tunning the bam file and base calibration. after all these preliminary steps i did variant calling by unified genotyper. still i have not used variant quality score recalibration pipeline.
I compared the result with the output generated for doing all these step on the single bamfile
i am surprised to see that the merged bam is showing a much larger number then the single bamfile. Am i doing right or is it because of error
e.g in single file chr1 has 2472 variants while in merged the number is 21853
just to mention
five bam files were from single sample run in six different lanes
Any help will be highly appreciated

[vivek_]

Did you check if GATK is treating the bam files as from a single sample or is it treating them as 6 different samples?

[huma Asif]

how can i find that ???
what i can understand from vcf output it is taking each lane as separate sample

please suggest where i am wrong

[kmcarr]

please suggest where i am wrong

When you created your individual read groups you defined each one as coming from a different sample.

@RG ID:XP3_NoUVA PL:SOLID PU:1_1 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane1
@RG ID:XP3_NoUVA.1 PL:SOLID PU:1_2 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane2
@RG ID:XP3_NoUVA.2 PL:SOLID PU:1_3 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane3
@RG ID:XP3_NoUVA.3 PL:SOLID PU:1_4 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane4
@RG ID:XP3_NoUVA.4 PL:SOLID PU:1_5 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane5
@RG ID:XP3_NoUVA.5 PL:SOLID PU:1_6 LB:XP3_NoUVA PI:0 DS:75x35RR DT:2013-04-12T08:00:07-0300 SM:Free_Exomes_april2013_lane6

If it is indeed the same sample run across several lanes then the sample definition for all read groups should be identical, e.g. SM:Free_Exomes_april2013. The lane identification is already included in the Platform Unit [PU:] tag for the read group.

[vivek_]

If re-doing the merge step is more computationally expensive, you could modify the BAM header and use samtools re-header to affix the new header to the bam before variant calling.

[huma Asif]

Thank you guys for your help
it fixed the problem