Difference between revisions of "WormBase-Caltech Weekly Calls"
From WormBaseWiki
Jump to navigationJump to searchm (→March 1, 2017) |
|||
| Line 22: | Line 22: | ||
| − | == March | + | == March 2, 2017 == |
=== Expression Cluster SVM === | === Expression Cluster SVM === | ||
Revision as of 18:53, 2 March 2017
Contents
Previous Years
2017 Meetings
March 2, 2017
Expression Cluster SVM
- Started curating expression clusters in 2007
- Had been using text pattern match to identify expression cluster papers
- Now with more papers for positive and negative training set, shifting to SVM identification
- 540 positive papers, 900 negative papers for training
- Wen received first batch, roughly ~70% precision, possibly ~80-90% recall
- Wen reviewed papers from WormEXP (http://wormexp.zoologie.uni-kiel.de/wormexp/)
Karen's SBIR grant
- Looks like it will get funded
- Karen will need to become part time at WormBase (needs to be more than half time on SBIR)
- Paul put in a request for posting another curator position
- If anyone has ideas about a potential (part-time or full-time) curator, let Paul know
- Will need to consider re-allocation of curation responsibilities
- Molecule curation in maintenance mode, may require updating of mapping terms with our internal IDs
- Ranjana: want to incorporate molecules in disease curation for WB/AGR, can offer to take over maintenance
- Do other AGR members curate molecules? In some cases yes; don't capture metabolites
- Transgene curation could be distributed
- Karen will be working on more automated curation of transgenes from papers; maybe also for molecules
Metabolomics
- What would we like to see in WB for metabolomics?
- Marian Walhout lab created the WormFlux database (http://wormflux.umassmed.edu/)
- WB enzyme gene pages link out to respective gene pages on WormFlux
- How are enzymatic pathways represented in GO, or by phenotype info? What's the cross-talk?
Pathway representation
- What are the best ways to capture empirical evidence as support for pathway models?
- Want clear, granular evidence for each assertion in a model, but need to accommodate probabilistic statements
- How can micropublications be used to fill in the gaps?
SObA GO on human data
- What would be a good set of genes to test?
- If we omit IEAs (most from mouse), are we left with substantial data for human genes?
