bamToFastq not working with .bam file from bwa-mem

[prs321]

I have aligned my paired end reads using bwa mem. I then filtered my .bam file in order to obtain a .bam file that has all the unmapped reads.

When I tried to take this .bam file and convert it into 2 .fq reads for each replicate, i kept getting this error:

*****WARNING: Query M00532:8:000000000-A17VF:1:2114:24702:5749 is marked as paired, but it's mate does not occur next to it in your .bam file. Skipping.


What do I do?

[DeeDeeBSE]

Which bamToFastq tool are you using?

[prs321]

bedtools bamToFastq

[GenoMax]

When you filtered the data did you keep the other read if only one out of the pair was mapped? Looks like you did not do that.

[prs321]

Here is an example:

M00532:8:000000000-A17VF:1:2110:8760:26938 121 NODE_140_length_92945_cov_15.514477 91194 60 28S101M126S = 91194 0 AATGTCGCCTCGGTCCGCCGACTCCCAGCCGCACCTACCTGACCGATGAACAGATCAATGGCGACTGTGCGGCGCTGCTGACGGCGCTGGCTTCGCGGTTGACCCCTGGCGAGAAGGTGAACGCCGAGTCTCCTTCCTCCTTCCCTCCCTCCTTTCCTCCTTCTCAGTGCGCACTTCGCCAGCATTCCTTCTCTTACCCCTTACTCTCTGACCCCTTCCACCCCGCGCTGCCGCAGGACGGGCTGTTGGTGCGCT //''-''-''''''.'.-.((((((.''-'<6/(.....(/(6=;;6.((/(EEEE?96<46.(6;2'45';5/<66/-'..5'2<<;6.-'''64/(6/((;6<;?269/((.6(669;42?6(///((((((((((((((''(('((.((((((-((/(((((-.'-'6/(1))@E@9*110***********<3+++++2++++++33)+43+5))55**5+5+CC5CAEC>>>CC@@@@@<@@<+===<5, NM:i:4 MD:Z:5A7G49G0G36 AS:i:81 XS:i:0 RG:Z:cof1 SA:Z:NODE_140_length_92945_cov_15.514477,91388,-,222S33M,60,1;
M00532:8:000000000-A17VF:1:2110:8760:26938 2169 NODE_140_length_92945_cov_15.514477 91388 60 222H33M = 91388 0 CCGCGCTGCCGCAGGACGGGCTGTTGGTGCGCT *5+5+CC5CAEC>>>CC@@@@@<@@<+===<5, NM:i:1 MD:Z:32G0 AS:i:32 XS:i:0 RG:Z:cof1 SA:Z:NODE_140_length_92945_cov_15.514477,91194,-,28S101M126S,60,4;
M00532:8:000000000-A17VF:1:2110:8760:26938 181 NODE_140_length_92945_cov_15.514477 91194 0 * = 91194 0 TGCTGCAGCGGATTACGCTTCACCGTCTGCCGTGCAGTTCGCAATTTCTCCACACGGCCCCGCCAGTGCATGCGGCCTGTTCCGGGTTTCCTACCTGCCCTGCCCCAACCCGTCCCCCGTCTCCGGGTGCCACTCCGTCGGATCCGCTTAACTCCTGGTGGATTTGTTGACCGCCCCCTGCTTCCGCGGTTGCTTGTCCGCCTGAATCCCCCCCCCTTGGGTTTTGGTTTTCCCCTCCACGCCGTCGTCCGCCA (((.''((((((((/'((((((('(('''.(/((/(''-(((((/(((((((''44'-'''/(.((((-('''(((((.'-'-(((;(/((.('(((''6-(''.'(.'-'.'4.'-((-'''.(/((((-(.-'.'9/..'''.(6./(/((((((((((.//(/.(-''.''-'((((('''-''(((6/(('--4..((((((-.'''5)<95+9E5+5++C5--5++*+5++>55<5555+55==<<=<, AS:i:0 XS:i:0 RG:Z:cof1



Here is the command line i used after mapping my reads in order to filter:


samtools view -u -f 4 –F 264 alignments.bam > tmps1.bam
samtools view -u -f 8 -F 260 alignments.bam > tmps2.bam
samtools view -u -f 12 -F 256 alignments.bam > tmps3.bam
samtools merge –u output.bam input1.bam input2.bam input3.bam
samtools sort –n input.bam output

bamToFastq –i unmapped.bam –fq unmapped_reads1.fastq –fq2 unmapped_reads2.fastq

[prs321]

bump still need help

[prs321]

bump still need help

[prs321]

bump still need help

[prs321]

bump still need help

[dpryan]

I don't know that bamToFastq supports chimeric alignments, such as the one you showed.

[prs321]

I don't know that bamToFastq supports chimeric alignments, such as the one you showed.

When I used novoalign and then converted my bam file into the unmapped fastq files, the unmapped fastq files did not contain any chimeric reads.

But the BWA one does. Is it better to just to leave these reads out?

[dpryan]

Given that novoalign at least used to not produce chimeric alignments, that's not surprising. You can just filter out the reads with the 0x800 bit set in the flag and then things should work.

[prs321]

So essentially get rid of the chimeric alignments? How do I "filter" them using BWA? or what program will I end up using?

[dpryan]

samtools view -bF 0x800 foo.bam > foo.filtered.bam