search

nunofonseca / irap

integrated RNA-seq Analysis Pipeline
GNU General Public License v3.0
82 stars 33 forks source link

Kallisto: handle fragment length better #68

Closed pinin4fjords closed 6 years ago

pinin4fjords commented 6 years ago

Currently iRAP is specifying read length for the fragment length parameter in Kallisto (-l), in both single- and paired-end read in irap_quant.mk, and for single-end in the other usage in irap_map.mk.

Read length is unlikely to be a good estimator of fragment length. For paired-end libraries it would be better to allow Kallisto to make its own estimates, while for single-end a fixed typical value of ~200 is probably more sensible than using read length, in the absence of better estimates from experimentalists.

nunofonseca commented 6 years ago

Thanks for the PR and sorry for the delay. To speed up things I replaced this PR with the following commit https://github.com/nunofonseca/irap/commit/23278491ab75e3e1604329812b6a41bf1b37aa8f . The commit avoids a hardcoded value for fragment length by using variables (kallisto_se_fragment_length and kallisto_se_sd); enables semantics of this PR when atlas SOPs are used; by default keeps backward compatibility. Cheers.