BWA MEM question

[CHObot]

I am using BWA MEM to map 250 bp MiSeq PE reads. I am doing targeted genomic sequencing, enriching the library for specific transgenic sequences. I make a reference sequence, map all of the reads to that reference (using the default parameters). What I get are many improperly mapped reads. Some of the reads that map have 20 - 50 mismatches between the read and the reference, but it maps them none-the-less. There are similar sequences contaminating the sample apparently, but I don't want these spurious mappings. How can I increase the stringency of the mapping, so I don't get all of these imporperly mapped reads? I have more than enough coverage to throw out any bad mappings and still have tons of coverage.

Thanks!
CHObot

P.S. I am looking for chimeric reads at the ends of the mappings to determine transgenic insertion sites. Is BWA MEM the best algorithm for this? It seems to work, albeit with a lot of manual examination of chimeric reads.

[sdriscoll]

you should be able to threshold the alignments by the MAPQ value. I'm not sure of the specific cutoff but let's say 20.

samtools view -bq 20 alignments.bam > filtered.bam

bwa has a very descriptive MAPQ field so this *should* work. Optionally maybe there's another tool out there that can specifically filter by mismatches if you want to keep the mismatches down to say 4 or 5% of the read length.

[jp.]

can some post general commands to align fq PE with hg19
where should I start with ?

[mastal]

the bwa manual webpage gives examples of the various bwa commands:

http://bio-bwa.sourceforge.net/bwa.shtml

[jp.]

I used bwa aln -t 4 -f testsample.sai /GATKbundle/ucsc.hg19.fasta /testsample.fastq
It was finished within 2 minutes and generated testsample.sai. Is this the alignment file ?
I used bowtie and it took 4-5 hrs but bwa is so fast ??? something is wrong ????

the bwa manual webpage gives examples of the various bwa commands:

http://bio-bwa.sourceforge.net/bwa.shtml

[mastal]

bwa aln is part of a 2-step process.

the first step generates the .sai file.

the second step is bwa samse/sampe, depending on whether you have single end or paired-end reads, and should give you a .sam file.

the manual will give you the details.