HLA typing with bwakit

[evakoe]

Dear all,

has somebody attempted HLA typing with bwakit and can share some experience?
I have whole exome sequencing data of human individuals for who I would like to do HLA typing. I ran the bwakit as instructed in the Readme, but have problems interpreting the output. Here are the first few lines from the out.hla.all for HLA-B for one individual.

HLA-B*50:01:01  HLA-B*55:01:01  0       0       2
HLA-B*50:01:02  HLA-B*55:01:01  0       0       2
HLA-B*50:01:04  HLA-B*55:01:01  0       0       2
HLA-B*50:01:01  HLA-B*55:01:03  0       0       2
HLA-B*50:01:02  HLA-B*55:01:03  0       0       2
HLA-B*50:01:04  HLA-B*55:01:03  0       0       2
HLA-B*50:01:01  HLA-B*55:01:07  0       0       2
HLA-B*50:01:02  HLA-B*55:01:07  0       0       2
HLA-B*50:01:04  HLA-B*55:01:07  0       0       2

Here Heng Li explains the meaning of columns 3 to 5:
column 3: #mismatches on the primary exon(s)
column 4: #mismatches on all considered exons
column 5: #exons used in typing

My questions are:
1. What are columns 1 and 2? I would have expected results for one allele, not two.
2. What is the "best" or the "true" allele? The out.hla.top file simply gives the first line of the out.hla.all for each gene, but from the evidence in the out.hla.all file, the first line does not seem to be better than the second or third.
3. Would you say that given high coverage exome sequencing data, one should be able to clearly identify (with bwakit) the true allele an individual has for each of the HLA genes?

Thank you
Eva

[evakoe]

For the people interested in this post:
1. I assume the first two columns are the HLA alleles for the two human alleles, so the HLA-B allele is heterozygous in this case.
2. The coverage was probably not high enough to determine a six digit resolution. After all, the first four digits are allthe same, and the last two are variable. I ran OptiType on the same sample which gave the output

B1	B2
B*50:01	B*55:01

3. The coverage on this sample is about 25X, which apparently is enough for four digit resolution with bwakit, but not for six digit resolution.

[afadda]

hi evakoe,

which command did you use? cuz i'm trying to do the same but it's not working.
i'm using this:
run-bwamem -o NA12878-hs38 -t 20 -H -s /gpfs/data_jrnas1/ref_data/Hsapiens/hs38DH/hs38DH.fa NA12878_2.fastq.gz NA12878_1.fastq.gz |sh
and it doesn't give me sorted bam or HLA typing.

[evakoe]

The command I used was

run-bwamem -t 4 -R "@RG\tID:myID\tSM:mySM\tPL:myPL\tLB:myLB\tPU:myPU" -H -d -s -o sample BWAindex sample_1.fastq sample_2.fastq | sh

and it ran as expected