interpretation of alignment results

[carolW]

Hi,
How to interpret the results of a program alignment like BWA or bowtie? In general, is it the percentage of different reads, for ex less than 2% or 5%? How about in details?

Any tutorial, review and documentation is welcome.

Look forward to your reply,

Carol

[swbarnes2]

No one can answer this without knowing what your experiment is. And only you know that.

[carolW]

As a newbie, I have started with the public data before analysing private data. So how to interpret already on the public data like 1000 genomes and then, focus on the private data?

[dpryan]

To reiterate what swbarnes2 said, you have to ask a coherent question to receive a useful answer. "Analyze" is a broad term, you need to specify what sorts of things you're interested in looking at. Given that you mentioned 1000 genomes data, are you interested in looking at SNPs?

[carolW]

yes, but already after the alignment, how can I evaluate that the results are good? for ex, I got the following result for one of the 1000 genomes after running bwa aln. I can think max of 9 mis-alignments could be considered as negligeable but how to evaluate in other cases? Is it 9 out of 225 bp read or number of seq 109811 should be considered?

[bwa_aln] 17bp reads: max_diff = 2
[bwa_aln] 38bp reads: max_diff = 3
[bwa_aln] 64bp reads: max_diff = 4
[bwa_aln] 93bp reads: max_diff = 5
[bwa_aln] 124bp reads: max_diff = 6
[bwa_aln] 157bp reads: max_diff = 7
[bwa_aln] 190bp reads: max_diff = 8
[bwa_aln] 225bp reads: max_diff = 9
[bwa_aln_core] 109811 sequences have been processed.

[Inma]

I think it's a good idea visualizing data from BWA in IGV (http://www.broadinstitute.org/igv/) or tablet (http://bioinf.scri.ac.uk/tablet/) and check the sequence coverage or the way your reads are mapping in the genome.

[carolW]

I think it's a good idea visualizing data from BWA in IGV (http://www.broadinstitute.org/igv/) or tablet (http://bioinf.scri.ac.uk/tablet/) and check the sequence coverage or the way your reads are mapping in the genome.

Thanks for your input.

I can do this as the data set is small but would it be a good idea to do so if the data set is large?

[Inma]

I did this in a 5 Mb bacterial genome, mapping 301453 reads against the sequence. I don't know if this programs are restricted to a small set of data.

[carolW]

And then, how to evaluate quantitatively the sequence coverage checking and acceptable number of mis-alignments?

[swbarnes2]

And then, how to evaluate quantitatively the sequence coverage checking and acceptable number of mis-alignments?

You can get an estimate of coverage by looking at the # of aligned reads, length of reads, versus the length of your genome.

There is no magical value to tell you if reads mis-aligned. If reads mis-aligned, it's because your genome is repetitive, or it doesn't match what you actually sequenced. Neither one is the fault of the software, and neither one is something you really need to worry about much, let alone try to use as a general guide to how well your experimetn worked.

What is more helpful is to know the % of reads that aligned to your target. If the benchwork people screwed up the library prep, or misinformed you as to what the sample actually is, an abnormal alignment % will tell you that.

What % is expected depends on what the experiment is, and how good the people doing the benchwork are.