Multimapper-aware bigwig tracks

[emattei]

Hi, I am interested in generating bigwig tracks for the RNA that I am quantifying using TElocal. How can I build a track that takes into account the EM read assignment from TElocal? Do you have a script/suggestions?

Thank you

[olivertam]

Hi,

Thank you for your interest in the software.

Unfortunately, that is a non-trivial idea, as you would need to know which read was redistributed, as well as the potential fractional count to the locus. This would then require the extraction of the particular alignment for that redistributed read, and providing a fractional count before generating your bedGraph/bigWig.

I will confer with my colleagues to see if they have other ideas/suggestions, but currently this is outside the scope of the TElocal software.

Thanks.

[emattei]

Can this information be outputted by TElocal? At some point while the algorithm is running it is available. I understand that this is more a future development

[olivertam]

Hi,

It would be quite difficult as the EM doesn't assign the read to a single annotation, but rather proportionally based on EM, so you could get multiple annotations per read that you would have to disentangle afterwards. Furthermore, the current code removes the read ID and the alignment information once the initial weights are assigned. Thus it would require a major code rewrite to both retain the read ID, and also the relevant alignment information for the proportional distribution, and would definitely result in a significant increase in memory usage.

This is definitely something that could be considered, but will take quite a while to implement properly.

Thanks.