WormBase-Caltech Weekly Calls September 2019

September 12, 2019

Update on SVM pipeline

• New SVM pipeline: more analysis and more parameter tuning
• avoiding precision (and F-value) as a measure (dependent on ratio of positives and negatives in test set)
• For example shown, "dumb" machine starts out with precision above 0.6
• G-value (Michael's invention); does not depend on distribution of sets
• Applied to various data types
• Analysis: 10-fold cross validation
  • Randomly select 10% pos and neg (without replacement) and repeat until all papers sampled
• F-value changes over different p/n values; G-value does not (essentially flat)
• Area Under the Curve (AUC): probability that a random positive scores higher than random negative
• AUC values for many WB data types upper 80%'s into 90%'s
• Ranjana: How many papers for a good training set? Michael: we don't know yet
• Can't reproduce old training sets (for old SVM); provide Michael better training sets if you want improved SVM
• If SVM still not good enough, Michael will work on deep neural networks (Tensor Flow)
• Michael can provide training sets he has used recently

Clarifying definitions of "defective" and "deficient" for phenotypes

• WB phenotype ontology has many "variant/abnormal" terms and distinct subclass terms for "defective/deficient"
• Have tried to create a logical definition pattern for these terms, but the vagueness of the meaning of "defective" and how it is distinct from "abnormal" has stalled the process
• What do we mean exactly by "defective" and how, specifically, is this distinct from "abnormal"?
• Definitions include meanings or words:
  • "Variations in the ability"
  • "aberrant"
  • "defect"
  • "defective"
  • "defects"
  • "deficiency"
  • "deficient"
  • "disrupted"
  • "impaired"
  • "incompetent"
  • "ineffective"
  • "perturbation that disrupts"
  • Failure to execute the characteristic response = abnormal?
  • abnormal
  • abnormality leading to specific outcomes
  • fail to exhibit the same taxis behavior = abnormal?
  • failure
  • failure OR delayed
  • failure, slower OR late
  • failure/abnormal
  • reduced
  • slower

Citace upload

• • Tuesday, Sep 24th

Strain to ID mapping

• Waiting on Hinxton to send strain ID mapping file?
• Hopefully we can all get that well before the upload deadline
• Will do global replacement at time of citace upload (at least for now)

New name server

• When will this officially go live?
• Will we now be able to request strain IDs through the server? Yes

SObA Graphs

• New graphs now live on site (Expression, Gene Ontology, Human Diseases, Phenotypes)
• A lot of whitespace padding above and below graph; maybe trim? trimming vertically would ultimately limit the view pane when user wants to zoom in, so we should leave as is for now
• Diff tool: Raymond and Juancarlos created a prototype diff tool (for comparing two genes, for example)
  • Paul: compared two genes that should be very similar, but there are a lot of differences; may reflect annotation coverage rather than biology

September 19, 2019

Strains

• Need to wait for new strain IDs from Hinxton before running dumping scripts
• Don't edit multi-ontology strain fields in OA for now!
• Juancarlos will map free text and ontology-name strain entries to strain IDs once we have the complete mapping file
• "Requested strain" field in Disease OA; not dumped, so don't need to worry about right now

Alliance literature curation

• Working group will be formed soon
• Will work out general common pipelines for literature curation

SObA Graph relations

• Currently only integrating over "is a", "part of" and "regulates"
• Maybe we could provide users an option to specify which relations to include, or maybe just exclude "regulates"

Author First Pass

• Putting together paper for AFP
• Reviewing all user input for paper
• Asking individual curators to check input

September 26, 2019

Data mining

• Someone in Paul's lab asking to retrieve list of C. elegans orthologs from a list of human genes
• Could we build a (simple) Alliance tool to do this?
• Could SimpleMine do this? Could we build a SimpleMine-like tool for Alliance?

Strains

• Paul D generated WBStrains for the missing TransgeneOme objects
• Working on a pipeline to identify new TransgeneOme strains at each upload
• One TransgeneOme object had 2 strains. Possible solutions: dump 2 expression objects that differ only in the Strain or remove the UNIQUE tag in the data model
  • Probably best to keep UNIQUE tag
• Raymond: concerned about automatically generating strains based on imports from the group
• Many odd strain names are coming from the TransgeneOme group; maybe we ought to have more discussions about generating official (following nomenclature standards) strain names from their imports
• Quarantine strains on initial import; review and accept if pass standards

Community phenotype requests August 2019

• Sent out new round of phenotype requests on August 20, 21, and 22, 2019
• 2,626 emails/papers requested
• 114 emails bounced; 5 resent to new addresses
• 460 Phenotype OA community annotations; 181 RNAi OA annotations (641 annotations total)
• From 94 papers (83 for Phenotype OA; 33 for RNAi; 22 for both)
• By 81 distinct community curators (70 for Phenotype OA; 32 for RNAi OA; 21 for both)
• 50 papers flagged as not having phenotypes (40 papers DO have phenotypes; 10 marked as negative; 80% failure rate!)
  • Email states: "If there are no nematode phenotypes in this paper click the following link :"
  • Maybe people are confused, or want to blow off the request
  • Maybe we can programmatically generate short URLs for the link? May be difficult
  • Provide a link to correct mistakes on confirmation page
• 4 papers flagged for phenotypes (only 2 had curatable phenotypes; 1 had honey-induced phenotypes)
• 115 papers with responses (5% response); 24 papers with input that were not main focus of request
• Can we provide an opt-out link?

Comparison SObA

• Actually quite complicated; may require more consideration

SObA graph and Ontology Browser for papers

• May be able to modify/hack existing tools for genes and apply to papers
• Paper-term matching powered by Textpresso