2007 Ontology Action items

From GO Wiki
Jump to: navigation, search

Ontology Content – Midori and David

Ontology development group communication – Wiki, email, regular virtual meetings.

Midori refers us to development group report for specific progress metrics.


Highlights

Immunology terms, CNS development, GO-CL cross products, MIT collaboration, biological process is_a complete, regulation (cross products and regulation).

• GO-CL cross products involve creating cross product definitions in the GO with the cell ontology using a yet to be determined set of new relationship types from the relations ontology – these cross products are initially proposed by OBOL, vetted by humans, and results are fed back to OBOL to improve OBOL guesses – eventually these proposed cross products will be integrated into the GO – project was put on hold as attention was directed to other efforts, but cross-product effort is almost done.

• MIT group including Jonathan Liu, Gil Alterovitz has done information entropy analysis on GO terms, relationships and annotations. This group has identified terms that are “too general” and “too specific” and need to be moved. The “too specific” recommendations usually specify grouping terms that need to stay in place, but the “too general” recommendations seem to be picking out real problem areas of the ontology where some more specificity (new terms) is needed.


David discusses is_a complete in biological process


About 700 terms in biological process were is_a orphans.

Big part of the effort involved making sure everything in biological process is actually a biological process – fair amount of graph reorganization was necessary.

Group met almost every day for an hour or so via Skype and Webex.

In general, non-is_a-complete terms weren't very good terms.

One problem was discrepancy between organismal processes and cellular processes – solved problem by introducing cellular processes, multi-organism processes, multi-cellular organismal processes (these three are disjoint). This helps solves the old problem of single cells being whole organisms.

Metabolism became metabolic process – homeostasis broken into homeostasis of cells, homeostasis of chemicals.

Added developmental processes (which includes any process in development).

All top level categories are is_a processes, then the children of these terms can be part_of.

Still a few outliers like “photosynthetic water oxidation” that couldn't be cleanly placed, and are left as direct children of biological process.

Biological regulation is now a broken into regulation of molecular function, biological process, biological quality – all of the GO regulation terms should be decomposable into cross products.

David wants tools to make it impossible to add non-is_a-complete terms to the GO .


Action item for John – make it impossible to add non-is_a complete terms in OBO-Edit