WormBase-Caltech Weekly Calls October 2011

October 6, 2011

SVM

• SVM Manuscript in process of resubmission
• Messy procedures in place; needs to be cleaned up
• Yuling able to get pipeline running from beginning to the end
• May take longer than previously expected to get up and running
• FlyBase has code running for SVM
• Yuling should contact FlyBase to make sure they're OK, once he has WormBase SVM working
• SGD would like to incorporate SVM into they're database
• Yuri would work on SGD SVM, ultimately

Physical interaction model

• Close to done
• Once model approved, work on:
  • Converting BioGRID file to the ACE model
  • Convert existing YH data into new model

Image curation

• Need to establish legal process for acquiring images
• Contact Caltech Office of the General Counsel for approval

WORM Paper

• Karen working on it
• Need a clearer/new plan

Expression pattern display

• Showing composite images with the option to see greater detail
• Separate out detail images according to Qualifier tags (Certain, Uncertain, Partial)
• Render oblique angles of cells/tissues to show 3D features/morphology of anatomy objects and, possibly, animations
• "Real" images should always take precedence to "virtual" images
• May talk to John Murray or Bill Mohler for confocal images of embryos

Worm Method (Worm Book)

• Any new web pages for C. elegans researchers, send to Raymond
• URL-finder for neuroscience (Yuling)

Textpresso

• Working on Fly Textpresso site (and others)
• Rose Oughtred would like Wnt Textpresso tools
• PDF tools (URL extractor)
• SVM re-working
• Arun working on GSA editor

Grants

• Need to think about next grant in a couple of months (winter); due in October 2012? 30 pages, short?
• Focus on writing papers (anatomy function, WORM, virtual worm, etc.)
• Focus on human-relevance
• Need Gantt charts and quarterly statements of progress
• Stats
  • Where are we with each data type
  • Status of website
  • Automation
  • What is different in the field?
  • How is data changing?

October 13, 2011

WORM Publiction

• Looks good overall
• Change "library" of papers to "index"
• Putative deadline of November 15th
• Put in more details about the new website?
• Figures? Screenshots from new website? Gene page features?
• Save other (detailed) content for future papers

Paper pipeline

• Need to fix some PubMed IDs
• Over 900 articles in Postgres do not have a PubMed ID; Why? How to fix?
• Could some aspects of the pipeline be accomplished by a non-PhD?
• Supplemental hires? Students?
• What tasks would need to be done?
• Can these tasks be scripted/automated? Community input?

SVM

• Almost all scripts working
• Output of SVM slightly different from Ruihua's results
• Need to resolve the differences
• Data inconsistencies
• Need to spot check the papers to see what the major problems might be
• Yuling ~90% confident in his results
• Get feedback from relevant/interested curator(s)
• One-day CPU/computation time (~30 hours for 70 papers with 9 models)

NAR Publication Accepted

Genetic Interaction Curation

• Sharing curated genetic interactions with BioGRID
• What tools will each database use and how will we share data?
  • Both BioGRID and WormBase will use the IMS tool at BioGRID for physical interactions
  • WormBase will use the Ontology Annotator for genetic interactions
  • BioGRID may still use IMS for genetic interactions, but the format will have to be parsed to populate Postgres
• BioGRID-curated genetic interactions can be flagged as such for later review

Physical Interaction model

• Kimberly and Paul Davis discussing
• XREF questions
  • Paul would like to minimize the number of XREFs in the model
  • Which XREFs can we remove?

Elsevier Linking

• Establishing pipeline for linking Science Direct papers to Wormbase and viceversa

October 20, 2011

Data Submission Working Group

• One representative from each site
• Raymond from Caltech, with Juancarlos
• Have a discussion for every type
  • How are we doing?
  • Are we up to date?
  • Do we want unpublished data?
  • Quality control
  • Data type priority?
• Groups/people that have a lot of data of a particular type; acquire their data
  • How can we facilitate data submission from these groups/individuals?
  • Ultimately like to train these users/submitters to use curator tools
• Form-filling as part of publication process?
• What data types are we missing? Wiki Pages
  • qPCR?
  • Nanostring?
  • Single molecule studies, absolute quantities?
  • SAGE?
  • 3C (Chromosome Conformation Capture), 4C, 5C
  • Metabalome?
  • Pathways/Processes?
  • Drug/disease interactions?
  • Infections?
  • Examples of C. elegans as a model?
• Expression data
  • How do annotate a presence/absence call on expression of a given gene?
  • rpkm cutoff?
  • Number of molecules?
  • Case-by-case thresholds
• Disease vs Process pages?
  • 100 diseases modeled in C. elegans?
  • Flat-file capture and display of disease-relevance information (as opposed to ontology-based)
• "Subject" vs. "Process" vs. "Disease"
  • Examples of "Subject": Sewage treatment, Biofilm formation, etc.
  • Examples of "Disease": Alzheimer's, Huntington's, Cancer, etc.
  • Examples of "Process": Development, Innate immunity, Organ formation, etc.
  • User experience?
• User-submitted data inconsistency/discrepancy flag?

WORM Paper

• Still requesting any major comments or content changes
• "Final" version available next week

C. elegans Research Resources Chapter (WormBook)

• Yahoo-like Yellow pages of useful URLs with brief descriptions of each site
• URLs being pulled from papers by Yuling

October 27, 2011

NAR Proofs are in

• Karen will send around
• Going back out tomorrow; everyone check
• Add Arun as author

Predicted Gene Interactions

• Wei wei contacted Paul
• Who is the contact person? Xiaodong can be point person
• Add URLs to Gene Orienteer and Interolog database on gene pages
• Issue is with syncing the databases and updating

SVM

• Yuling finsished running SVM pipeline; can make run every two weeks
• Need to learn how to best train the models; do we need to retrain?
• Gary Williams sent e-mail; incorrect paper IDs on some SVM results
  • Gene structure correction papers have been incorrect
  • May need to retrain the gene structure correction SVM
  • Need someone at Caltech to work with Yuling to retrain the model
• Kimberly will look through e-mail archive about the issue
• We could really improve the training set, as original set may have been inadequate
• What data types do we retrain: Expression pattern? Phenotype?
• Improve training by including verified positive papers from last two years (since last training)
• In past, needed 400 papers to train; would be best to lower requirement
• Antibody and transgene SVM works well; maybe incorporate for Expression pattern SVM?
• Use pattern matching to correct SVM results or vice versa?
• Good to make progress on Gene Structure correction papers
• Keeping track of sentences/paragraphs; performing SVM on paragraphs (vs papers)?

Collecting URLs from papers (Raymond)

• For interesting/relevant URLs for C. elegans
• Journal URLs are problematic
• Yuling runs scripts on Textpresso corpus to collect URLs
• PubMed XML of journals (Kimberly)

WORM Publication

• Final version available
• Review for corrections
• To submit early next week (before Wednesday)