Reliability

[Layla]

I am comparing the output from Bowtie and maq and using the command below I get a total of 147-e6 alignments for 34-e6 reads. After rigorous filtering, I have half the numbers of reads mapped to chr1 as I did using maq. One of the issues is this:
HWI-EAS261:6:61:525:1199#0/2 - gi|89161185|ref|NC_000001.9|NC_000001 78 TAACCCTAACCCTAACCCTAACCCTAACCCAACCCTAACCCTAAC
HWI-EAS261:6:61:525:1199#0/2 + gi|89161205|ref|NC_000003.10|NC_000003 199384671 GTTAGGGTTAGGGTTGGGTTAGGGTTAGGGTTAGGGTTAGGGTTA

Maq flags the second alignment with an MAQ score<10 and is thus easy to eliminate maintaining the 1st alignment.
Is it possible to know which alignment is correct from Bowtie?

./bowtie -t h_sapiens_asm -1 sanger1.fq -2 sanger2.fq -n 2 -l 28 -e 70 -I 0 -X 250 -a > output.txt --un unmapped_reads.fq --al reads_at_least_1_alignment.txt
#Reported 73,823,285 paired-end alignments to 1 output stream(s)

Layla

[Ben Langmead]

Hi Layla,

If filtering of repetitive alignments is what you're after, take a look at the -m and --strata options. Together, these options allow you to define what hits count as "repetitive", then filter them out. The --max option additionally diverts the repetitively-aligned reads to a separate output file.

Also, I can't see how you're running Maq, but you might be in a situation where you'd actually like to run Bowtie in unpaired mode separately on your two mate inputs. I say this because the Maq output you show seems to have aligned the mates to separate chromosomes. Perhaps I'm misinterpreting.

Thanks,
Ben

[Layla]

Hi Ben,

Thanks for your reply. I've got myself a little muddled with the output. I used the -m1 -k1 option but I was worried I might be missing reads that had 1MM or 2MM as these would go unreported. Hence I asked everything to be reported and will clean the data myself. I will re-align the unmapped reads using unpaired option. The above command for 34-e6 reads (17-e6 reads per file) took 6hours on a 32GB machine, is this reasonable?

I might get back to you once I have looked at the output some more.

Thanks,
L

[Layla]

Quick question,

How does Bowtie deal with paired end reads where both ends align to different chromosomes? I believe maq gives various flags to identify such reads.

Thanks,
L