Ruchira S. Datta
Ruchira Datta: Berkeley PHOG: PhyloFacts Orthology Group
I won't be able to talk and comment at the same time :-) - Ruchira S. Datta
superorthology: all nodes in the path should be speciation events - Albert Vilella
superorthology is transitive but highly restrictive (low coverage) - Albert Vilella
PHOG is based on precalculated trees in PhyloFacts: 55k protein families (and expanding) - Albert Vilella
allow users to run their own family in the system, gets incorporated - Albert Vilella
get more signal by including more taxa in the tree - Albert Vilella
PHOG-O standard definition of orthology - Albert Vilella
SCI-PHY a method that cuts the tree into functional subfamilies - Albert Vilella
PHOG search results: genes can belong to multiple orthology groups based on regions selected, or time points at which sequences were clustered - Albert Vilella
RNN orthologs -- reciprocal nearest neighbor - Albert Vilella
PHOG-T thresholded PHOG -- if the duplication distance is less than the threshold, keep proceeding up the tree to the next maximal node, e.g. some very recent duplications are allowed in the superorthology definition - Albert Vilella
Note to self: Ilan wants to know about the pipeline - Ruchira S. Datta
there is some discussion of benchmarks - Ruchira S. Datta
Toni asked about effect of incompletely sequenced genomes - Ruchira S. Datta
Erik asked, isn't evaluating specificity of tree-based methods based on a tree circular? I say it might be a fair point. Christophe says, the definition of orthology is based on the tree, so he thinks this is a fair assessment. - Ruchira S. Datta
someone says: too many dbs, European funding agencies should pick one - Ruchira S. Datta
someone asks: are these databases independent of each other? A: yes, they're all independently computationally derived - Ruchira S. Datta
David Roos: however, we all use overlapping sets of methods - Ruchira S. Datta
Erik: have you tried using confidence measures from the db? A: Not so much confidence in the confidence measures. Doesn't want to put spurious levels of confidence. David: This isn't an official confidence level. Julie: Can confidence measures be compared among different databases? David: That's not the use case. If have competing calls, could choose between them based on confidence measures. - Ruchira S. Datta
Christophe: What dbs provide confidence measures? Erik: InParanoid. Normally recommend people only use orthologs with confidence scores > 0.9. - Ruchira S. Datta