HLA region alignment

[jbelmont]

I have used BWA to align 2 complete genomes (32X average coverage, Illumina 100bp paired-end). Can anyone advise on the performance of BWA in the HLA region? Is it usual to have low alignment efficiency for chr6:28,400,000-33,500,00 (hg19)?

[Richard Finney]

Did you include any chr6 haplotype regions in your reference?
How about printing out your header information for your bamfile (use samtools -H parameter)?

[jbelmont]

HLA region alignment

chr6 references
from samtools view -H
@SQ SN:chr6 LN:171115067
@SQ SN:chr6_apd_hap1 LN:4622290
@SQ SN:chr6_cox_hap2 LN:4795371
@SQ SN:chr6_dbb_hap3 LN:4610396
@SQ SN:chr6_mann_hap4 LN:4683263
@SQ SN:chr6_mcf_hap5 LN:4833398
@SQ SN:chr6_qbl_hap6 LN:4611984
@SQ SN:chr6_ssto_hap7 LN:4928567

Are these adequate for alignment on HLA region?

[Richard Finney]

Adequate? More like overkill.
The haplotypes are alternate versions of the same territory.
You'll probably want to strip out the "haploytpes" unless you're sure about what you're doing and that's what you want. Keeping the "randoms" regions as part of your reference is okay; though many just drop these for convenience.
You can re-run without haplotypes or
1) Strip out the haplotypes reads
2) dump these reads from the haplotypes and re-align them to a "haplotype-less" reference.
3) then merge these back.

Repair or re-do.

You can also just leave it and remember to adjust for any automatic analyses done on this regions.

[ymc]

Well, bwa is not designed to map reads to highly polymorphic region. Your best bet is to try GSNAP which can take dbsnp data to turn some mismatches to matches.