WormBase-Caltech Weekly Calls February 2011
February 10, 2011
- Should we have a Caltech project/site manager? We'll look at the issues to be solved first
- Todd would like to know more about what's going on at Caltech
- What would Todd like to know about? New ideas, data types?
- What can we communicate more effectively to Todd?
- Should we take minutes of Caltech WormBase group meetings and send around?
- Kimberly: helpful (for off site individuals) if we go around the room to talk about what each of us is working on
- Bitbucket use
- How do we see only what pertains to us, individually?
- Paul: good way to document bugs and fixes, problems/solutions
- Bitbucket Wiki - does that capture what people want/need to see?
- Who should Todd follow? What would Todd like to know about?
- Raymond: used for code development/versioning by OICR; may be the best use for Bitbucket
- Should we develop best practices guideline for Bitbucket use?
- Avoid posting topics that are too specific or too vague?
- What do we want most to get out of Bitbucket?
- Curation efforts? No
- SAB Meeting
- Not much feedback for literature curators
- User interface stuff took precedence
- Paul: Testing, spot-checking website for errors
- Web issues, e-mails from Gary and Oliver
- WormBase-wide conference call twice per month
- First & third Thursday of the month
- Starting next Thursday (2-17) @ 8:30am PST
- Web redesign meetings on Thursdays will have to be every other week? Do both same day?
- Raymond: WormBase mirrors
- Do we want to implement the Genome Browser for the mirror(s)? If practical
- It seems as though Caltech mirror has been crashing often
- Has the Caltech mirror been working for people lately? Yes, more or less
February 17, 2011
Interaction dumping script
- Spot check table
- Push changes to Git? Yes, to main branch
Interactions automatically downloaded to FTP every release? Yes
- People should check their own curation data
- Users query frequently?
- Can WormMart work instead?
- Ruihua - feature can be added
- Todd - good to have pre-defined queries generated automatically
- Parse/remove predicted interactions (400,000) from others during dump? Yes
- Paul - Wei Wei wants to help with updates - who should she contact at WB?
Ranjana - Solar flares causing static on phone? ;)
Newsletter for new website release? Yes
- Todd - 5 development milestones
- teleconference with Gary Ruvkun's lab next week
- Go live with Beta version in June (@IWM)
- Go live live with new site in September
- Retire old site at end of year
- Need outreach, documentation, screencasts
- WormBase to BioGRID curation
- Issues of mapping interaction types: WB -> BioGRID vs. BioGRID -> WB
- Moving towards using WB interaction types
- Definitions of our interaction types
- Rose will propose to BioGRID community
- Physical interactions - BioGRID has a better format
- break physical interactions away from other interaction types?
- Kimberly can propose changes to model: YH separate from other interactions?
- Physical vs Genetic interactions? keep separate
- We need an interaction ontology, of some sort
- Who should Wei Wei contact, how to coordinate with Hinxton; getting data from FlyBase and SGD?
- Interaction pages - explain what the interaction is based on; provenance/reference
- Gene Orienteer shows data, but user has to link out
WormBase IWM souvenir?
- Computer stress ball? with new site address?
- bouncy ball?
- worm rubber band thing?
- Screen shield?
- Complementary iPads? ;)
- Antifungal socks? wristbands?
- new data doesn't load; problem runnning search engine; can load it on athena, not generalizable problem?
- OS was not complete with an update; try again
- two virtual machines working back and forth for support
- how often is SPELL being used? log?; had 6-7 users within two weeks complaining of SPELL problems
- Todd - can send Google analytic tool to put at bottom of page
- Wen - SPELL testing server; official and mirror; want a separate testing server for new releases; mirror on athena (Wen's working machine); Raymond says don't want to attract users to that machine; WB's running SPELL on altair; heavily loaded and has had problems; relying on athena may be problematic
- maybe we need analysis of all of our machines to see big picture; can we consolidate machines; make use of ones we have? If not we can get another machine.
- what type of machine do we need?
- Log of all machines with purchase date and functionality?
- Linux vs Mac curators? athena development machine;
- how much power do the scripts need? can check
- may not be appropriate as official server if main machine goes down
- WB and athena can run SPELL server, but not others. security problems?
- prefer work machine is readily re-bootable; reconfigure software easily without affecting other things that people rely on
- need 10GB space; 4GB memory? SPELL production server; want it stable for outside world; hardware designed for that purpose; farm it out to IMSS?
- complexity of the application; Linux can run multiple applications; caprica has SVM production and WormMart; understand demand of the application; requirements; development vs running scripts, how often? Efficient use of machines. Have 2 computers, getting a third; for WormMart need power; traffic suddenly increased yesterday; monitor traffic; not much personal experience with server maintenance; cluster needs lots of expertise
- building official sever of SPELL at OICR? Yes
- with WormMart and WormBase, people just want to know if it's down, just to know to lessen frustration; Official statement from dev team
- WormMart - we give users testing url; can we just change WormBase link to that testing server?
- going to change HTML to update users with most recent information; discuss with Lincoln, 5 datasets on testing server; remaining 3 datasets. if WormMart not working, put a message on main site
- it doesn't look good to have a link to a tool linked from a production server; if we don't trust the data we shouldn't put it up.
- would like to keep the html page as a testing server; could modify page
- if WormMart is working OK but not perfect, that's OK; Comment in the banner
- the html page is a useful tool to get feedback from users
February 24, 2011
CIT Computer Survey - done
CD/DVD burner not working (Wen)
- run @ OICR
- dynamic/static files
- problem: no one takes care of them
- curators should claim them and take care of them
- Gene interaction dump script claimed by Raymond and Xiaodong
- 256 interaction objects empty (should not be); emptied during build?
- Monitor usage with new site
- Files (like FASTA, GFF) should be immediately available to users
- Page that links to the scripts (FTP)
- Need to define what we should have (common datasets)
- Example: Table with all genes, RNAi phenotype, genetic phenotype, etc.
- How many genes have been knocked out by RNAi?, etc.
- Metrics table on each species: #genes, #chromosomes, etc.
- Curators think about what data should be available
- Modifying existing script is easier than making new one
- Gary Williams already generates RNAi-phenotype-gene connections?
Snehal work half-time at WormBase?
- Will interview
- UI testing?
- Curation? GO? Phenotype?
- Start April 15th - 8 months? (Sept.) maybe longer
- Focus on particular papers?
Physical Interaction model
- Have been discussing with Rose at BioGRID
- Y2H, CoIPs, Pull downs, etc.
- Store physical interactions in WB outside of GO curation
- Data model
- YH model vs Interaction model
- Physical interaction tag in current interaction model unused
- Generalize YH model -> physical interaction model
- Add experiment types
- Issue: what do we do with existing physical interaction tag in interaction model?
- Split out different interaction types into individual classes?
- Separate into physical, genetic, predicted classes? Textpresso?
- Create separate classes for each experimental type? (Y2H, CoIP, etc.)
- XREF Interaction to Physical interaction, Genetic interaction
- Goal is for cleanliness and completeness
- How to capture experimental-specific info in curation and database?
- Suppress Y2H import from BioGRID?
- Keep interaction models as they are?
- All curators: Review interaction models to see what they want on the curation end