by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Rik
    @blahah
    tomorrow I'll describe what I've got so far for contigs
    Rob Patro
    @rob-p
    Yes, I agree --- there's no reason not to discuss everything here
    since (1) it's in the open and everyone can see it and contribute thoughts and (2) any good (and also bad ;P) ideas are recorded
    Avi Srivastava
    @k3yavi
    oh by that I didn’t mean like not discussing here
    It’s more like how should I really start looking into the read based problem.
    Like @Blahah said earlier
    we need to take into account the substitution errors
    in reads. i’ve started looking into that
    but didn’t get much success.
    Avi Srivastava
    @k3yavi
    hi @cboursnell @Blahah ,
    hope you guys are doing good,
    Thanks for your twitter posts.
    we're trying to put on some more applications for quasi-mapping
    and are looking for assembled and simulated data.
    @Blahah, I have plant data which you have given me some time back, but
    the problem with that is # of transcript wrt to # of gene are almost same in rice (Orzya sativa ssp. japonica) data.
    If you guys can point out to some reference it'll really help.
    Rob Patro
    @rob-p
    Hi guys --- just to chime in here, the problem we're facing is that in this assembly, there are very few contigs. Further, in the reference annotation, the number of transcripts is almost equal to the number of genes (~97k transcripts and ~91k genes). We are curious if you know where we might find an assembly that has a dense txp <-> gene mapping (e.g. mouse, human, chicken, etc.)
    Rik
    @blahah
    Drosophila has a lot of alternative splicing
    Rik
    @blahah
    Hi
    Avi Srivastava
    @k3yavi
    hello
    Is there any way we can get assembled data of Drosophila somewhere
    Rik
    @blahah
    Ensembl has it
    Sorry on mobile and have baby in one arm so msgs are short!
    Avi Srivastava
    @k3yavi
    :smile: not a problem
    Rik
    @blahah
    2w2w
    Eh... Accidental typing
    Avi Srivastava
    @k3yavi
    Actually, we’d need a De-novo assembly of organism with alot of alternative splicing
    Rob Patro
    @rob-p
    So we're trying to do an experiment similar to that in Corset
    so, to do the clustering, we need the transcripts to be expressed (which a de novo assembly gives us)
    we could also do these experiments on a reference (which we're currently looking into), but then we miss out on the peculiarities of "real" de novo assembled data
    Rik
    @blahah
    Why not just use the corset data? I contacted them after the paper came out and got them to release the data on Zenodo
    Rob Patro
    @rob-p
    did not know that
    Rik
    @blahah
    We used it in the transrate paper
    Rob Patro
    @rob-p
    that sounds great!
    Rik
    @blahah
    On mobile but if you search Zenodo it should come up
    sweet!
    thanks :)
    Rik
    @blahah
    Np! Real credit goes to Nadia and
    ... Alicia
    Avi Srivastava
    @k3yavi
    sorry, wrong post.
    Chris Boursnell
    @cboursnell
    How is it going on transcriptome clustering using rapmap? I've been playing with Heng Li's minimap recently and was wondering if you had anything similar?
    Rob Patro
    @rob-p
    It's coming along --- we're fleshing out a pipeline now. Our current approach is more akin to Corset (group transcripts by multimapping reads) than e.g. Vsearch (directly compare all transcripts). However the initial prototype gave promising results (see the "clustering" section in the RapMap pre-print).
    The thing we're looking at now is how to work "backward" to extract gene models from our clusters (and improve the clustering in the process)
    Rob Patro
    @rob-p
    @cboursnell --- how was minimap? seems really cool
    Chris Boursnell
    @cboursnell
    Very fast :P I'm hoping that I can use the output to give something similar to the vsearch clustering output to be able to replace that in transfuse. Not sure quite how to do that yet... maybe some sort of agglomerative hierarchical clustering...
    Rob Patro
    @rob-p
    We've had good luck, so far, with graph-based clustering
    Chris Boursnell
    @cboursnell
    I applied MCL to the output from minimap. I'm fiddling with some settings at the moment... I'm not sure this method is going to work out for what we want for transfuse though.