A great tool to count multiple aligned reads.

[a.kmg]

Dear members of seqanswers,

Do you know a tool for counting multiple aligned reads?

Currently I use the htseq-count tool, but it doesn't take into account this type of read.

I know cufflinks allows this kind of thing, but I do not like this tool, because for me it is a "black box".

There are some time ERANGE allowed to do this assigning multiple aligned reads in the proportion of single aligned reads in each respective region. I find it effective, unfortunately ERANGE is no longer maintained.

I thank you in advance for your help.

Best,
a.kmg

[dpryan]

The appropriate tool will depend on how you want the reads counted. featureCounts is one option, though obviously if you're counting over genes and using those integer counts to look at differential expression then this is a bad idea. Other options would be things like eXpress or any other tool meant to generate expected counts using a expectation maximization method. This will yield fractional counts, which is appropriate for multimapped reads.

[a.kmg]

Thank you dpryan.

I currently testing Rcount, a new tool where alignments of multireads are weighted based on the number of uniquely aligned reads in the neighborhood.

And I tested featureCount, that actually allows me to treat multiple aligned reads, but the fact of counting the reads for each chromosomal locations on which it aligns leads to a biais, right?

[dpryan]

And I tested featureCount, that actually allows me to treat multiple aligned reads, but the fact of counting the reads for each chromosomal locations on which it aligns leads to a biais, right?

Not really. The reads arose from the neighboring reads, so the counting would accurately represent the biological reality.