Bowtie Paired end 100% failed to align

[rahul.m.dhodapkar]

Hello all,

I am trying to use Bowtie to align paired end sequencing data from solexa, but I am consistently getting

# reads processed: 3188
# reads with at least one reported alignment: 0 (0.00%)
# reads that failed to align: 3188 (100.00%)

I have tried running Bowtie's single end mapping, and have gotten both ends to align with >90% of the reads with at least one reported alignment, but the second I try paired-end, the whole thing fails, and there are no error messages to give me some sort of hint as to what is happening.

./bowtie -S -t -p 8 -q --chunkmbs 128 hg18_combined.fa.bowtie -1 Pair1.fastq -2 Pair2.fastq bowpeout.sam

is the command that I am using. Has anyone else run into this problem before? Does anyone have any ideas as to what I should do to get some of the reads to align properly?

-Rahul Dhodapkar

[john_mu]

Did you check if the reads are forward-reverse? or some other orientation?

[rahul.m.dhodapkar]

Re: orientation

john,

yes I have tried running bowtie with all of the orientation options that it provides, and I consistently get the no alignment problem. But thanks for the idea

-Rahul

[Ben Langmead]

john,

yes I have tried running bowtie with all of the orientation options that it provides, and I consistently get the no alignment problem. But thanks for the idea

-Rahul

Hi Rahul,

Are you sure the -I and -X options encompass (or mostly encompass) the distribution of insert sizes in your sequencing library?

Ben

[rahul.m.dhodapkar]

Ben,

Yes, I am expecting to have insert sizes of about 200-300, which is pretty close to the default bowtie -I and -X options. I would certainly expect that at least one read fall into that category. Just to be sure, I ran it again with the same command as above except that the -I and -X boundaries were set to 0 1000, and I still got no reads aligned.

-Rahul

[Ben Langmead]

Ben,

Yes, I am expecting to have insert sizes of about 200-300, which is pretty close to the default bowtie -I and -X options. I would certainly expect that at least one read fall into that category. Just to be sure, I ran it again with the same command as above except that the -I and -X boundaries were set to 0 1000, and I still got no reads aligned.

-Rahul

And when you align them as unpaired reads, the effective insert sizes you observe are in the 200-300 range?

Can you please post the full commands used to run Bowtie in unpaired and paired modes, and one example of a read that should align in paired-end mode but doesn't?

Thanks,
Ben

[rahul.m.dhodapkar]

GA-B_0003:5:95:3459:14691#0/1	0	chr4	357461	255	99M	*	0	0	AACATAAGAAAATTCATAGTGCAGATAAACCCTACAAATGTAAAGAATGTGACAAAGCCTTTAAACAATTTTCGCTCCTGAGTCAACATAAGAAAATTC	fffffffffffffffffefffffeffffffefefffffffeffffffffefcefffedeefffecffefffff^eeffdfYfdfcfeffeef\fddfdf	XA:i:0	MD:Z:99	NM:i:0

aligned on the pair1.fq

GA-B_0003:5:95:3459:14691#0/2	0	chr4	357538	255	99M	*	0	0	CTGAGTCAACATAAGAAAATTCATACTGTAGATAAACCCTACAAATGTAAAGATTGTGACAAAGCCTTTAAACGGTTCTCACACCTGAATAAACATAAG	ecbda_cdd`f\dfffdfeadeeedddeaefffffbcdddfeffffffffcffffedfefffffdfffffffeffffffffffffffffffffffffff	XA:i:0	MD:Z:99	NM:i:0

aligned on pair2.fq

the exact commands that I used were

./bowtie -S -t -p 8 -q --chunkmbs 128 hg18_combined.fa.bowtie pair1.fastq pair1.sam

for the single end (then repeated for pair 2 with the exact same syntax.

./bowtie -S -t -p 8 -q -I 0 -X 300 --chunkmbs 128 hg18_combined.fa.bowtie -1 Pair1.fastq -2 Pair2.fastq bowpeout.sam

for the paired end mode output. hope this helps

-Rahul Dhodapkar

[kmcarr]

Rahul,

The SAM output shows that both of the reads are aligning to the forward strand (flag==0), but for Illumina paired end reads, if read 1 aligns to the forward strand then read 2 should align to the reverse. Also, the quality string for read 2 looks as if it has been reversed (poorer scores at the 5' end, improving towards the 3' end).

Did you (or someone else) reverse complement the reads in Pair2.fastq prior to running them through Bowtie? Bowtie wants the reads as they were produced by the instrument. If one of the input files has been reverse complemented the aligned pairs won't be considered correct by Bowtie.

[rahul.m.dhodapkar]

kmcarr,

working from your observation, I tried to run the bowtie paired end alignment command with the --ff option on, and it worked! Thank you all so much for your help

Gratefully,
Rahul

[OTU]

Hi All!

I am facing some kind of the same problem, but all posted possible solutions are not working...
I tried --ff, --fr, --rf - and no result. 0% of alignment. I even tried -U option - no result.

Does anybody know what may be the reason why??

[kmcarr]

Hi All!

I am facing some kind of the same problem, but all posted possible solutions are not working...
I tried --ff, --fr, --rf - and no result. 0% of alignment. I even tried -U option - no result.

Does anybody know what may be the reason why??

Maybe because the reads you are trying to align aren't really from reference genome species? What did you sequence? What reference genome are you using?

[OTU]

The reads are from metagenome, and the reference is an indexed bowtie library of Silva SSU rRNA genes from all living organisms.
So... I t should give some hits!

[kmcarr]

The reads are from metagenome, and the reference is an indexed bowtie library of Silva SSU rRNA genes from all living organisms.
So... I t should give some hits!

What does "The reads are from a metagenome" mean? Is it shotgun DNA-Seq from a bacterial community or did you do create 16S amplicons and sequence those?

[OTU]

Sorry for confusion!

So, sequences are from sorted cells from water sample, and should be enriched in one candidate division sequences. Cells were lysed and amplified via multiple displacement amplification. After - sequenced on MySeq platform.

[kmcarr]

...Silva SSU rRNA genes from all living organisms.
So... I t should give some hits!

Well, Silva does not in fact have SSU genes from all living organisms.

sequences are from sorted cells from water sample, and should be enriched in one candidate division sequences

What candidate division would that be? Could it be something not cataloged in the Silva database?