MSA skype 20100727

Back to


Mark, Kyle, Andreas

Weekly Update

Mark: Lots of reading this week and a whole bunch of little improvements. Cleaned up things a bit and removed compiler warnings.

Memory issues

Mark: Tested larger Pfam families. Turns out the Piwi family was just big enough to fit into memory. (1.5GB) Snoal and Susd are too large to fit into memory.

Current results: 800 sequences in Susd (278.5 average residues in sequences). Raw sequences 619k in mysql Snoal: shorter sequences (168.6 average residue nr): 1000 sequences

Seems all of the pairwise alignments are still in memory when they are not needed any more.

We will try to improve memory consumption

Structure alignments

Andreas: Where are we regarding adding of structure alignment info as additional input? Mark: Came also up during reading on variable gap penalties. There are a number of possibilities how to deal with this. (mentioned several options) A: many options, which one to take? K: Simplest! M: Take the variable gap penalties and leave which constraints are chose as a dimension of the alignment step.

Only 1 month left we should make sure we only do things that are possible within this timeframe.

Outlook for this week

Most important for the moment is to improve memory consumption

Pfam quality benchmarking

Linear space algorithm Myers Miller