Infer transport reactions as biological processes about pathways2go HOT 8 CLOSED

regarding 'In this particular case, they’ve annotated it to the GO process term “regulation of canonical Wnt…”, which is a strange choice but you can still use it in the conversion.'

This actually is already used in the conversion, but not to assign a type to the reaction node. The reaction node is made 'part of' the process that Reactome tagged it with. You can see this happening in noctua-dev for the reaction 'Beta-catenin translocates to the nucleus' in the pathway 'Disassembly of the destruction complex and recruitment of AXIN to the membrane'

from pathways2go.

Should this cover the transport of complexes or just individual proteins? Would those result in different GO classifications?

from pathways2go.

Looking at OWL, making call that it should only apply to proteins. Otherwise need to find different GO classes to use.

from pathways2go.

Applied the OWL pattern described in the issue comment. Looks like it works nicely. For the test case 'beta catenin translocates to the nucleus', the converter code generates the base classification to 'establishment of protein localization' and adds the has_target_start_location and has_target_end_location properties. From there, Arachne or ELK successfully deepen the classification to 'protein localization to the nucleoplasm'.

from pathways2go.

All looking fine so far. Closing.

from pathways2go.

@goodb commented a long time ago, "In this particular case [“beta catenin translocates to the nucleus”, Reactome] ... annotated it to the GO process term “regulation of canonical Wnt…”, which is a strange choice but you can still use it in the conversion."
The reason for this assignment takes us off on a tangent from the central issue of this ticket. It happened because the transport reaction is part of the larger process of Wnt regulation and the Reactome curator did not associate a GO biological process term with the transport reaction. In a case like that (event instance has no BP term) the script that generates the Reactome GAF looks at its parent event and if the latter has a BP term it assigns that term to all proteins involved in the child event. Here that has resulted in the reasoning that as catenin transport is somehow part of Wnt regulation, it can get a Wnt regulation transport term.
The durable fix for this is for Reactome curators to become better at assigning BP terms and to assign them to leaf events, not only to higher-level processes. That's an enormous clean-up process. Perhaps a pilot-scale start would be to try to generate a list of strange choices of BP terms to look for patterns and also for cases that seem especially strange hence first targets for cleanup.

from pathways2go.

@deustp01 I'm trying to parse the above, is there anything wrong with the logic for this conversion as it stands? e.g. see the translocation reaction in http://noctua-dev.berkeleybop.org/editor/graph/gomodel:-1280847518

If I understand, things are working here on my end and will evolve naturally as Reactome makes changes upstream. And I'm assuming your are talking about the BioPAX export, not the GAF generation right? Otherwise I'm not quite understanding.

from pathways2go.

We need to talk this one through face to face, because it looks like there are several levels of mismatch / misunderstanding here (but also, I suspect, a good solution in that there are at a minimum some features of our annotation that can be used as red flags to tell you when to ignore one of our BP assertions).

from pathways2go.