search

cosmo-team / cosmo-issues

Issue repository for Cosmo (separate until we can transfer issues between repositories nicely - ignore code)
http://www.github.com/cosmo-team/cosmo
GNU General Public License v3.0
0 stars 0 forks source link

Address Bloom Filter Tries directly #147

Open alexbowe opened 8 years ago

alexbowe commented 8 years ago

This paper describes a a colored de-Bruijn graph representation for "pan-genome storage".

@christinaboucher said they can't handle the application we are considering. It might be worth saying that in the paper? Or we can just leave it out.

alexbowe commented 8 years ago

Temporarily assigning this to @christinaboucher (I think you might have the most familiarity with it currently) - we should make sure that it definitely can't do what we are doing... but I'm pretty sure it can? If you can query presence of kmers, then you can traverse and find outdegree by generating every possible target kmer. This should work with any way of representing a set of kmers and their colours...

/cc @simon-j-puglisi (in case you would like to pick this one up)

simon-j-puglisi commented 8 years ago

Are you saying we should include it in experiments?

On Mon, Mar 7, 2016 at 3:26 PM, Alex Bowe notifications@github.com wrote:

Temporarily assigning this to @christinaboucher https://github.com/christinaboucher (I think you might have the most familiarity with it currently) - we should make sure that it definitely can't do what we are doing... but I'm pretty sure it can? If you can query presence of kmers, then you can traverse and find outdegree by generating every possible target kmer. This should work with any way of representing a set of kmers and their colours...

/cc @simon-j-puglisi https://github.com/simon-j-puglisi (in case you would like to pick this one up)

— Reply to this email directly or view it on GitHub https://github.com/alexbowe/cosmo-private/issues/147#issuecomment-193247674 .

alexbowe commented 8 years ago

My main point is that our decision before (during a skype call, but my memory is imperfect) was to say that their structure doesn't support the required operations. However, their structure does support enough operations, but the effort involved may be non-trivial (although I haven't checked).

So, I'm saying we should decide how to address it:

simon-j-puglisi commented 8 years ago

Well the last point (Mention it, but don't do anything) is certainly the most appealing ;) We should really get to the bottom of it though (and compare if we have necessary go and time).

On Mon, Mar 14, 2016 at 8:57 AM, Alex Bowe notifications@github.com wrote:

My main point is that our decision before (during a skype call, but my memory is imperfect) was to say that their structure doesn't support the required operations. However, their structure does support enough operations, but the effort involved may be non-trivial (although I haven't checked).

So, I'm saying we should decide how to address it:

  • Ignore it
  • Implement the required code to compare them
  • Mention it, but say why we didn't compare them (perhaps the effort is non-trivial, but I don't think it is fair to say that they can't do it).

— Reply to this email directly or view it on GitHub https://github.com/alexbowe/cosmo-private/issues/147#issuecomment-196172761 .