Ballpark runtime figures – I took about 30 nuclear receptor genes and ran those against the 4000 MeSH Disease profiles – this took about 30 minutes. Extrapolating to the approximately 10000 human genes with GeneRIFs makes the runtime about 7 days.
Plan for tomorrow is to pull out an example for and mock up the distance [...]
Archive for the ‘todo’ Category
Towards Profile comparison results
Posted in progress, todo on April 22, 2008 | Leave a Comment »
Focus: MeSH to OMIM Disease Mapping
Posted in UMLS, todo on October 5, 2007 | Leave a Comment »
Important thing TODO, from today’s meeting: Create a validation set – Extract OMIM disease-gene links. Map OMIM genes to Entrez Gene. Map OMIM diseases to MeSH terms.
Of these, the OMIM-MeSH is likely the hard part. I should do this by UMLS, but UMLS is the ugliest database I’ve ever seen. Must no longer be afraid…
This Week’s TODO list
Posted in todo on August 27, 2007 | Leave a Comment »
Look at extracting TFs and brain diseases from OMIM
parsing OMIM, look for existing OMIM-to-MeSH binding
Do first order separation database query
sonoma back online, databases reloaded
write auto-reload database script
date stamp older versions, automatically invoke parsers as needed
skeleton of scripts and directories for rapid deployment on server
Back to the Research!
Posted in todo on August 10, 2007 | Leave a Comment »
Finally – it’s time to get back to research. A couple things currently on the plate as extra-curricular:
journal version of the n-SIFT paper. I should set up experiments to correlate angle of rotation with point match error
finish testing for the shape-based ChIP on chip expression peak modelling
Back to my actual research!
Speaking of [...]
Optimise
Posted in SQL, progress, todo on April 28, 2007 | Leave a Comment »
Sounds like a transformers yell…”Transformers, Optimise!”
Anyways, thing to fix: all the CREATE TABLE AS SELECT…. create tables without keys. No key is definitely bad. Trying this for the one that really matters – the gene_term_related_citations query. EXPLAIN seems happier, but proof will be in the pudding (or [...]
