Preparing files to call by Manta

Question

[German.M.Demidov]

Dear community members,

I face a problem while using Manta - it complains that 2 reads with the same ID exist in the read group.

Existing read: FCC3FB1ACXX:3:2313:16454:24691/2 chrom:pos:strand chr4:84367961:+ cigar: 53M47S templateSize: 53 SA: chr12,68316771,+,48H52M,122,0; mateChrom:pos:strand chr4:84367976:-
New read: FCC3FB1ACXX:3:2313:16454:24691/2 chrom:pos:strand chr12:68316771:- cigar: 33S67M templateSize: -67 SA: chr4,84367976,-,38M62H,62,0; mateChrom:pos:strand chr12:68316771:+

I decomposed the BAM file into FASTQs. Run Dragen from fastqs - same failure.

Could you help me, what should I do with my FASTQ files to avoid such reads that have same ID and are assigned to the same read group?

I wrote a script to keep only reads with unique IDs from FASTQs (decomposed from BAMs) - surprisingly there are tons of reads with same IDs and maybe just removal of equal IDs reads is not a strategy which can help.

[Asaf]

I think it has to do with the way the mapper keeps multi-mapping reads. Which mapper did you use?

[German.M.Demidov]

it is BWA for the initial BAMs and also BWA for re-mapping with Dragen (mem, I believe). These reads should be marked as secondary alignment - but they are not, flagstat tell me that I have only primary mapped reads.

Answer 1

[Asaf]

See Albert's point no 3 here: bwa mem and multiply mapped reads it's not secondary alignment, it's supplementary. In the example you gave, two parts of the read are mapped to different locations so both are primary but one is probably tagged as supplementary (bit 2048 as opposed to bit 256)

[German.M.Demidov]

Thanks a lot for your help! I understood the principle. Could you guide me which bit flag indicates a supplementary alignment? I guess it is not always 2048...

[German.M.Demidov]

What criteria are used to determine supplementary alignment with bwa-mem (flag 2048) sorry looks like it is always this flag =) your answer was very helpful! If you post it is an answer, I'll accept it.