How to fix bwa and ssaha2 misalignments?

[ymc]

At the 98th and 99th codon of HLA-DQA1 gene in the reference genome, NA12878 should be ATCATG (reference) and AGTCTG respectively for the two chromomsomes. However, when I looked at the bams (downloaded from 1000g) in IGV, I noticed that both bwa and ssaha2 added an insertion and a deletion to the alignment:

A-TCATG
AGTC-TG

How should I fix this? Increase the gap penalty and re-run the alignment? Is there a way to manually edit the alignment inside bam?

[ymc]

Fixed this by manually edit the sam file. Anyone know if this can be fixed by adjusting the gap penalty?

[ymc]

I tried GSNAP with SNP-tolerant on. However, I find that many reads trimmed this part for unknown reasons such that the coverage on the 98th and 99th codon in the alternate allele is very low.

[ymc]

Ran bwa again with gap opening penalty increases from 11 to 15. But I still have those insertion-deletion alignments. What should I try next?

[gaffa]

You might try running a local realignment program such as SRMA.

[nilshomer]

What's more likely, three SNPs in a row, or two 1bp indels and two matches?

The inequality you want to satisfy 3*-MM < 2*M + (-O + -E).
M = 1
MM = -3
O = -5
E = -2

I use these as the default parameters in TMAP.

[ymc]

What's more likely, three SNPs in a row, or two 1bp indels and two matches?

The inequality you want to satisfy 3*-MM < 2*M + (-O + -E).
M = 1
MM = -3
O = -5
E = -2

I use these as the default parameters in TMAP.

By default, bwa aln has M=1, MM=-3, O=-11, E=-4. It seems to me the inequality should be in favor of having three mismatches but in practice it is not. What's going on here?

[ymc]

You might try running a local realignment program such as SRMA.

Is this an abandoned project? The binary is dated 2010-10-22

[darked89]

Is this an abandoned project? The binary is dated 2010-10-22

Source on github seems to be newer, but still 1 year old:

https://github.com/nh13/SRMA

[ymc]

You might try running a local realignment program such as SRMA.

Oh well, I am getting hundreds of this arraylist error after 145min of srma run....

at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)
at java.util.ArrayList$SubList.add(ArrayList.java:965)

real 145m13.649s
user 177m44.719s
sys 3m20.097s

[ymc]

Source on github seems to be newer, but still 1 year old:

https://github.com/nh13/SRMA

Thanks. I am trying this slightly newer 0.1.16 version now

[ymc]

SRMA 0.1.16 also crashes

I tried "bwa aln -M 2" as well but the problem remains.

[darked89]

SRMA 0.1.16 also crashes

I tried "bwa aln -M 2" as well but the problem remains.

Hard to tell why. Few ideas:

1) Just in case: you are using sorted and indexed BAM as SRMA input?

2) Try to get some small test BAM file which is known to run OK with SRMA and then check if your local SRMA works on them.

3) if 1 and 2 are OK, something else may be wrong with your BAM. You may resort it with newest picard / validate the BAM.

4) if all else fails, switch to GATK

http://www.broadinstitute.org/gsa/wiki/index.php/Local_realignment_around_indels

[Genomics101]

I had the same problem. Calling the SNPs without BAQ in samtools mpileup fixed the problem (-B option). I have had the best results using the Extended BAQ, brings back the false negatives while reducing false positives as standard BAQ should.

[ymc]

Tried the latest GATK's indel realigner but it does seem to do anything to my problem

time java -jar GenomeAnalysisTK-2.0-31-gf57127e/GenomeAnalysisTK.jar -nt 6 -T RealignerTargetCreator -R ../exome/human_g1k_v37.fasta -o SRR098401_bwa.intervals -I ../NA12878/SRR098401_bwa.bam -known ../exome/Mills_and_1000G_gold_standard.indels.b37.vcf
time java -jar GenomeAnalysisTK-2.0-31-gf57127e/GenomeAnalysisTK.jar -T IndelRealigner -R ../exome/human_g1k_v37.fasta -I ../NA12878/SRR098401_bwa.bam -targetIntervals SRR098401_bwa.intervals -known ../exome/Mills_and_1000G_gold_standard.indels.b37.vcf -o SRR098401_realigned.bam