How to unique bwa mem reads?

Question

[niu.shengyong]

I use bwa mem with my paired-ended files and mixed reference genome (concatenate by hg19 and mm10) as the following command:

  bwa mem -t 4 mixed_human_mouse.fa \
  rep1.R1.decomplex.fastq.gz \
  rep1.R2.decomplex.fastq.gz \
  | samtools view -bS - > p56.rep1.bam

However, I get mapping in different species (both human and mouse) or different chromosomes in the several reads. How can I get the unique reads (with just one chromosome and one species in the single read)? The mapping sam file is shown below: (Human5 means the chromosome 5 in hg19)

CGGCTATGCGTACTATTCTCTCCGCCTATCCT:M01581:1209:000000000-D3YJT:1:1102:15099:1447 129     Mouse10 72297312        15      44M     MouseM  2300    0      TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTCTCTTTTTTTT    111>11110>0>///////>//>///<///<</0111<111/--    NM:i:0  MD:Z:44 AS:i:44 XS:i:40

ATTACTCGTAGGCATGCGTCTAATTTAGAGGC:M01581:1209:000000000-D3YJT:1:1102:15159:1448 81      Human4  31583390        60      44M     Mouse5  81894906       0       ATCCCATAAAAATATTTATCATTAGATTTAGCACATACCTGTAG    GA4FHHGHHHHHGGG5GFFBGFGFGFFGFGFCFFFFFFC3A3>3    NM:i:1  MD:Z:43T0      AS:i:43 XS:i:21

Simon

Answer 1

[GenoMax]

If you are interested in binning the reads into human/mouse (and mix) pool you should consider using bbsplit.sh from BBMap suite. see @Brian's answer in this thread: Tool to separate human and mouse rna seq reads

[niu.shengyong]

Thanks a lot! I'll look into it. How about the reads with different chromosomes?

[GenoMax]

You will have to decide what to do with the multi-mappers. BBMap give you multiple options (look at ambig=).

[niu.shengyong]

Thanks for your help! I successfully finish it. However, I don't know how to access to mapping statistics and check the quality. Do you have any suggestion?

[GenoMax]

BBtools produce stats in the STDERR by default. Did you capture that output? If not, you should be able to use reformat.sh on your BAM file to produce stats (histogram options). Otherwise Qualimap or samtools idxstats would work as well.