Hi
Patrick Aboyoun wrote:
Simon,
Could you provide some profiling information to show where the
bottlenecks are?
I don't know if there is really a clear bottleneck. 9 minutes to
calculate the coverage of 29 mio reads is 20 seconds per mio reads; this
is probably what the coverage function always needed. So, in the code
given in my mail, the summing up of the GenomeData objects is just
awkward but not a performance penalty.
> I am also wondering if I should be building up the
> functionality for RleList, which could have `+` and other Math
> operations. We have a lot of classes in the Sequence space and it is
> not clear yet which classes are going to be part of the winning
> solution.
I'd say that this is the main issue. I discover new classes every day.
You just mentioned 'RleList', Michael mentions 'GenomeDataList', and
Martin has another way to go again.
I'm sorry to say that, at least for me, this has become hopelessly
confusing, and I imagine that many other users fell the same. You write
that "it is not clear yet which classes are going to be part of the
winning solution" and I completely agree that it makes more sense to
have a few good classes rather than adding functionality to any class on
demand. So, maybe don't bother with a `+` operation for now.
Best regards
Simon
_______________________________________________
Bioc-sig-sequencing mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing
