I Need Some Explanation With Bowtie 2 (Xs And As)

Question

[veermyne]

hello everyone,

I juste started a project with Bowtie 2 and I need to compare the alignment score between the different possible alignments for a read (I'm using -k 15 ). I thought I could do this with the AS (and/or XS) value but these values are the same for all the possible alignment of a same read. So I guess I don't understand quite well how those two values work with the k15 option so I was wondering if one of you could help me by explaining what they mean or showing me a more exlpicit documentation than the bowtie 2 manual.

Thanks Gabriel.

[skymningen]

If there are 15 or more alignment possibilities giving the same score, you might end up with AS/XS being the same for all of the 15 alignments. As the alignment score only tells you, how close (or different) the aligned sequence is to the reference, this can occur.

[veermyne]

thanks for your reply, but on a second think it appears that I haven't been clear enough in my trouble. Fact is that the XS value is sometimes higher than the AS for all the possible alignment for a unic read. and none of the other alignment has these XS value as AS value.

So I don't understand: -how the XS can be better than all the AS (whereas the best score should be in the selected sequence, shouldn't it?) -if, as I understand that, the AS value is the score alignment for the current alignment, why do I have some alignment with a different XS and AS but no alignment corresponding to these XS score. (as they all have the same AS value why the XS value is not equal to the AS?)

thanks again

here an exemple of what I have:

HWI-1KL110:60:D1B59ACXX:6:1101:9407:13543    256    GRMZM2G117984_T02    599    1    35M    *    0    0    TACAAGATGGACAAGCAGGGGGCCAACCTTAGGTT    CCCFFFFFHHHHHJJJJJJJJJJJJJJJJJIJJFI    AS:i:55    XS:i:70    XN:i:0    XM:i:2    XO:i:0    XG:i:0    NM:i:2    MD:Z:3G25C5    YT:Z:UU

HWI-1KL110:60:D1B59ACXX:6:1101:9407:13543    256    GRMZM2G117984_T01    599    1    35M    *    0    0    TACAAGATGGACAAGCAGGGGGCCAACCTTAGGTT    CCCFFFFFHHHHHJJJJJJJJJJJJJJJJJIJJFI    AS:i:55    XS:i:70    XN:i:0    XM:i:2    XO:i:0    XG:i:0    NM:i:2    MD:Z:3G25C5    YT:Z:UU

here the XS >AS

[skymningen]

In the Bowtie2 Manual you can find this explanation:

XS:i:<n> Alignment score for the best-scoring alignment found other than the alignment reported. Can be negative. Can be greater than 0 in --local mode (but not in --end-to-end mode). Only present if the SAM record is for an aligned read and more than one alignment was found for the read. Note that, when the read is part of a concordantly-aligned pair, this score could be greater than AS:i.

[veermyne]

I saw this before but I'm not using paired end for the RNAseq. Only single end.

[Devon Ryan]

Just an observation (I don't use local alignment often), but the XS score in your example is equal to the maximum possible alignment score for those reads (35bp * default batch bonus of 2). I imagine that this sort of situation would always occur when bowtie2 finds more than 15 matches, since it might not be returning the optimal match in that case.

[veermyne]

that's an interesting observation, as far as I can see on my SAM this situation include the maximal alignment score as XS value. the trouble is that in the preceding example Bowtie 2 show only those two alignments so it's not an unshown alignment as their isn't 15 match already shown.

[Devon Ryan]

Interesting. So do you also see this behavior in reads having fewer than 15 alignments where the reported alignments themselves have multiple alignment scores (if that occurs)? This might be one of those cases where going through the bowtie2 code is the easiest way to figure out exactly why this happens.

BTW, why are you using -k 15 and then trying to determine the optimal alignment? If you only need the optimal alignment then just use the default mode.

[veermyne]

I'm using the K -15 mode because I want to know the reads which can have more than one alignment with the same alignment score and which have only one localisation.