WormBase-Caltech Weekly Calls
From WormBaseWiki
2012 Meetings
March 1, 2012
GO Meeting
- Focused on annotation pipelines; improving efficiency/effectiveness
- How to make GO annotations more 'expressive'
- GO would like to move towards more expressive statements
- Example: If a gene is involved in a function or process, where in the cell does this take place
- Common Annotation Framework
- Current/future members of the GO network can annotate using the same version of GO, same tools and standards
- Quality controls checks: e.g. do you have all the fields necessary to make an annotation
- GO hopes to centralize all of the data handling, formatting
- LEGO - Logical Extensions of GO
- We should pilot how we want to handle this; similar to how concise descriptions are constructed
- WormBase curates phenotypes, pathways, etc.
- Defining useful relationships to curate/annotate: Cross-products with defined relations
- Pilot: Take subdomains, pathways, try extended version of curation on these
- How do we capture that fly eye development is relevant to human biology?
- Humans don't have compound eyes - not the point
- The pathways are the same or similar; EGF signaling
- WormBase Process curation could really benefit from GO's adoption of this strategy
- Need to consider what the "right" way to approach this issue; need good pilot
- Where is the value? How do we focus on this?
- Another annotation pipeline: Phylogenetic Annotation and INference Tool (PAINT)
- How best to make these inferences?
- What kind of inferences can you make about organismal- or organ-specific processes?
- Uberon has framework for interspecies anatomical comparisons
- PAINT tool for nematodes?
Upload for WS231
- Interaction file upload took several hours
- Check if virtual memory is being used
- Likely culprit is the extra data and XREFs in the Interactor_info hash
- Can objectify the Interactor_info to be a tag in the main ?Interaction model
- We should warn EBI/Hinxton about this
WormBase Curator Interview next Thursday
Migration of Reporter_gene object annotations from Expr_pattern OA to Transgene OA
- Everything seems OK
SPELL
- Papers with less than three experiments, statistics calculations cause slow-down, memory limitations
- Now can bypass this problem
- We are now operating SPELL on our local machines
- Do Amazon instances function/behave differently than local server?
- Need to compare; find benefits & drawbacks
- Use Amazon server as a dynamic name server
- Users shouldn't notice a difference
- We won't need to ask Todd for anything; we can fix it ourselves
GO Meeting breakout session
- Software architecture for upcoming GO expansion (CAT - Common Annotation Tool)
- How does Textpresso integrate?
- What kind of annotation would GO expect Textpresso to do?
- User will be able to do guided text mining operations
- Example: regular expressions, then HMM, then export to CAT
- No forseeable roadblocks
- Maybe standardize all of the text mining types and methods behind them
- Develop paper-viewer? Apart from CAT, text mining flow? Separate module
March 15, 2012
RNAi clone mappings
- Can we submit PCR primers to Hinxton for genome mapping?
- They say so, we'll have to test
- Usually we deal with PCR_product objects like sjj_* or mv_*
Transgene sequences
- Authors explain PCR construction of transgenes; provide explicit sequence?
Should we make a standard submission format for authors?
- Maybe, authors should have to submit explicit sequences for RNAi probes, transgenes, etc.
- Enforce standards at what level? Editors, reviewers, journal, user community, etc...
- Submit PCR primers?
- We will draft a letter to community/editors to request explicit sequences for RNAi experiments
Should we perform sequencing of Ahringer (and other) clones ourselves?
- Write grant, perform sequencing?
New website release
- We will draft an e-mail/blog post about official release of new website at end of March
SVM Precision/Recall
- Standardize methods of calculating precision and recall on SVM results?
- True Positive = a positive identified by SVM that is actually positive
- False Positive = a positive identified by SVM that is actually negative
- True Negative = a negative identified by SVM that is actually negative
- False Negative = a negative identified by SVM that is actually positive
- Precision Rate (Positive Predictive Value) - how many SVM-identified positives are true positives? = True Positives/(True Positives + False Positives)
- Recall Rate (Sensitivity) = True Positives/(True Positives + False Negatives)
- Specificity = True Negatives/(True Negatives + False Positives)
- NPV (Negative Predictive Value) = True Negatives/(True Negatives + False Negatives)
Curator Candidates
March 22, 2012
New website release
- Sending around mass e-mail to WormBase community (~9500 e-mail addresses) announcing release