Does Picard tool support multi-fasta file?

[ikarus97]

Hi,

I'm trying to check alignment accuracy of bam files generated by tophat using Picard tools.
But the resulting mismatch rates ('PF_MISMATCH_RATE') were unreasonably high.
When trying to figure out why, I found some strange things.

Here is the entries in the sample bam file:

@HD     VN:1.0  SO:coordinate
@SQ     SN:chr1 LN:249250621
@PG     ID:TopHat       VN:1.4.0
1925_202_1426   0       chr1    1102502 255     23M     *       0       0       GCCATCTTACTGGGCAGCATTGG ___abbb_[NRb``[[`Y[_^&& NM:i:0  NH:i:1
49_971_1359     0       chr1    1102503 255     23M     *       0       0       CCATCTTACTGGGCAGCATTGGA QQIFIQ?@NW`VP@E[EGUVVEE NM:i:0  NH:i:1
76_239_469      0       chr1    1102503 255     22M     *       0       0       CCATCTTACTGGGCAGCATTGG  WWVTUXRS[^^`[CGL<N8;KK  NM:i:0  NH:i:1

All 3 reads were single-ended, and originated from chr1.fa (and aligned to chr1 perfectly).

First I tried 'CollectAlignmentSummaryMetrics' module just using reference sequence for chr1, and got mismatch rate of 0.

java -jar ~/picard-tools-1.67/CollectAlignmentSummaryMetrics.jar INPUT=hh.sam R=chr1.fa OUTPUT=tmp

But when I tried the same command with multi-fasta file containing every chromosomes, I got the mismatch rate of 0.77!

Is there anybody who can explain me what is wrong?
My first hunch is that this tool might not support multi-fasta format as reference. But then, how can I get the mismatch rate of a bam file?

[ikarus97]

I've got my answer: the reference sequences in fasta file should be in the same order as in the bam header.

[jh313]

Hey ikarus97, I was getting a similar issue. How did you fix the order of the reference sequences in the fasta file?