Noctua MOD Imports: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
mNo edit summary
mNo edit summary
Line 1: Line 1:
'''Note: the rules for MOD imports into Noctua are still under revision.  If you have any questions about specific rules, please enter a ticket in https://github.com/geneontology/gocamgen before doing any annotation revision.
'''Note: the rules for MOD imports into Noctua are still under revision.  If you have any questions about specific rules, please enter a ticket in https://github.com/geneontology/gocamgen before doing any annotation revision.'''


= Introduction =
= Introduction =

Revision as of 03:01, 11 April 2019

Note: the rules for MOD imports into Noctua are still under revision. If you have any questions about specific rules, please enter a ticket in https://github.com/geneontology/gocamgen before doing any annotation revision.

Introduction

This page contains documentation for importing a set of MOD annotations into Noctua as GO-CAMs (Gene Ontology Causal Activity Modeling). This initial set of guidelines is based on a set of general import rules that have been further refined by testing the imports with annotations from MGI and WormBase. We expect that each group that imports their annotations may have some unique issues that will need to be dealt with on an individual basis, but these guidelines are nevertheless intended to give all groups a point from which to begin the process.

Import Files

Annotation Sets Imported

  • MODs will import their manual GO annotations, e.g Assigned_by 'MGI'.
    • At this time, annotations made to the same species by external groups will not be imported.
    • MODs may choose to include/exclude annotations made with specific evidence codes, but the overall goal is to get a full set of manual annotations imported as GO-CAMs.

Import File Format

  • The annotation import pipeline uses the GPAD annotation file format.
  • Using GPAD file format allows us to import key information not in the GAF:
      • all gene product-to-term (gp2term) relations
      • annotation metadata in the Annotation_property field
  • The GPAD file used for annotation import may likely be a 'one-off' file that includes metadata stored in other curation tools that would not normally be output in the GPAD file consumed by our users.

Modeling Conventional Annotations as GO-CAMs

One Gene = One GO-CAM

  • Each gene's set of conventional annotations will be imported as one GO-CAM.
  • Annotations with no extensions are modeled very simply, while annotations with extensions are converted to a GO-CAM model that most faithfully represents the information captured in the extension.
    • Note that not all relations used in annotation extensions are used in Noctua and GO-CAMs, so the exact translation of an extension to a GO-CAM is not always possible. However, in these cases, we have tried very hard to capture the information in a clear, consistent way.

Basic Rules

Note: to check examples, search for the WGGene id on noctua-dev.berkeleybop,org

Molecular Function

  • [GP] <- enabled_by [MF]
    • Example: C. elegans atg-18 - WBGene00018294
  • [GP] contributes_to -> [MF]
    • Example: C. elegans pptr-1 - WBGene00012348
Protein Binding Annotations
  • Currently, GO-CAM models annotate protein binding differently than conventional annotation.
  • The difference is that the interacting partner(s) is captured as contextual information using the 'has input' relation.
  • In conventional annotation, the interacting partner is typically captured using the With/From field, although if the experimental interactor is not from the same species, the biologically relevant interactor may also have been captured using the has_input relation.
  • Proposals for Importing Protein Binding Annotations (assuming single-organism processes)
    • Import annotations 'as is'
      • Protein binding annotations would be imported 'as is', i.e. interacting partners would be captured in the With/From field. If there is also a 'has input'/'has direct input' extension that would be imported as contextual information.
      • Annotations would need to be processed after importation to comply with GO-CAM annotation guidelines.
      • Pros: minimal work upfront for annotation groups
      • Cons: Work would need to be done subsequent to import
    • Keep With/From data but also populate 'has input' with the same information
      • This approach would duplicate information in the With/From field in the 'has input' annotation extension
      • Would need to decide if this would happen pre- or post-import
      • Pros: This would be compliant with current GO-CAM annotation guidelines and would maintain current use of the With/From field
      • Cons: This would result in redundant information in the With/From vs extension fields, and could possibly result in incorrect modeling if the species in the With/From field is different from the primary species annotated.
    • Populate 'has input' only
      • This approach would remove the interacting partner from the With/From field and populate the extension.
      • Rule will be implemented as part of import
      • Pros: Compliant with current GO-CAM annotation practice
      • Cons: Not compliant with current conventional annotation practice
  • Points to consider:
    • GPAD exports - what should we export for protein binding to be consistent across the GOC?
    • Note that protein binding annotations in GO-CAM either describe the primary, or main, function of the gp or are used as evidence for a direct causal relationship between two activities.
      • If the results of protein binding experiments are used solely as evidence for a direct causal relationship, do we still want a protein binding annotation output in a GPAD and how would it be generated?
    • Should we take steps to change how we represent protein binding across the GOC (a proposal to do this has been out there for a long time)?

Cellular Component

  • [GP] part_of -> [CC]
    • Example: C. elegans epg-6 - WBGene00012641
  • [GP] colocalizes_with -> [CC]
    • Example: C. elegans mlc-4 - WBGene00003372

Biological Process

  • Biological Process annotations are modeled with respect to Molecular Function, as this is the GO-CAM convention.
    • All Molecular Functions will be to the root node; this is the most conservative approach.
    • Note that these root MF annotations will NOT be exported in the GPAD file.
  • Relations between the root MF and BP will be derived from existing gene product-to-Biological Process relations in the GPAD.
GP 'involved in' BP
  • These will be modeled as root MF 'enabled by' GP is 'part of' a BP.
    • [GP] <- 'enabled by' (GO:0003674 molecular_function) 'part_of'-> BP
    • Example: C. elegans ogt-1 - WBGene00003858
GP ' acts upstream of (or within)' BP
  • [GP] <-enabled_by (GO:0003674 molecular_function) causally_upstream_of_or_within -> BP
    • Example: mouse Rela - note that this rule is not implemented yet
    • Positive effect example:
    • Negative effect example:

References

  • Where multiple, pipe-separate references exist for a single annotation, only one reference ID is imported in order of preference:
    • PMID
    • GO_REF
    • doi
    • MOD paper id
  • Models when there is multiple evidence to support a given GO annotation:
    • If multiple, independent references exist to support a given annotation, do we want to combine them or leave them separate? What are the implications for later model manipulation?
    • Combining evidence on to a single edge wherever possible will simplify the graph display, but may make moving or merging annotations later more complicated.

With/From Field

  • Values in the With/From field will be imported as string literal, i.e. text will be imported "as is" with no further modeling in the OWL representation
  • github ticket on modeling multiple values in the With/From field in OWL, i.e. commas and/or pipes
  • Future project may be to develop and use OWL Union and Intersection to represent what is meant by commas and/or pipes in the With/From field

Annotation Extensions

Simple Conversions

  • These simple conversion rules are derived from a set of rules that Paul T. and Dustin put together.
  • Some existing annotations violate these rules, so it will be necessary to review those and make sure we're clear on what was meant by the annotation before we relax the rule.
Molecular Function
  • No more than one occurs_in(CC)
  • No more than one occurs_in(CL)
  • No more than one happens_during(BP)
  • No more than one part_of(BP)
  • No more than one activated_by(ChEBI)
  • No more than one inhibited_by(ChEBI)
  • No more than one occurs_in (UBERON)
  • No more than one occurs_in (EMAPA)

Multiple values allowed:

  • WBbt (contains both cell and anatomy terms)
  • has_input/has_direct_input(geneID or ChEBI)
    • has_direct_input will be converted to has_input upon import
Cellular Component
  • No more than one part_of(CC)
  • No more than one part_of(CL)
  • No more than one part_of(UBERON or EMAPA)
  • No more than one exists_during(GO:BP)
  • No more than on exists_during(WBls)

Multiple values allowed:

  • WBbt (contains both cell and anatomy terms)
Biological Process
  • No more than one occurs_in(CC)
  • No more than one occurs_in(CL)
  • No more than one part_of(BP)
  • No more than one happens_during(BP)
    • How many of these are really part_of(BP)?

Multiple entries allowed?:

  • occurs_in(UBERON)
  • occurs_in(EMAPA)

Multiple values allowed:

  • WBbt (contains both cell and anatomy terms)
  • has_input/has_direct_input(geneID or ChEBI)

Other Conversions

has regulation target
  • The rules below are still being tested and may be further refined.
regulation of molecular function
BP descendants
  • Annotated term is 'regulation of molecular function' (GO:0065009) or BP descendant
  • Expressed as [GP-A]<-enabled_by-[root MF]-part_of ->[regulation BP]-regulates -> [MF-Z] enabled_by ->[GP-B]
    • Note that MF Z should appear in the logical definition of the term “regulation of MF Z” (or in a triple with the regulates relation, so we can get the GO ID from that).
      • For example:
        • 'regulation of histone deacetylase activity' (GO:1901725) is defined as: 'biological_process' and (regulates some 'histone deacetylase activity')
  • The variations on this are all children of the term “regulation of molecular function” (GO:0065009).
  • Proposed directional variations are:
      • “positive regulation of MF Z” and children which would be expressed with the positively_regulates relation instead.
      • “negative regulation of MF Z” and children which would be expressed with the negatively_regulates relation instead.
MF descendants
  • Annotated term is 'molecular function regulator' (GO:0098772) or descendants
  • Generally expressed as [GP-A]<-enabled_by-[root MF]-regulates->[MF Z]-enabled_by->[GP-B]
  • Molecular function activators:
    • [GP-A] <- enabled_by - [root MF] - directly_positively_regulates -> [MF-Z] - enabled_by ->[GP=B]
  • Molecular function inhibitors:
    • [GP-A] <- enabled_by - [root MF] - directly_negatively_regulates -> [MF-Z] - enabled_by ->[GP=B]
  • Example: enzyme activator activity
    • molecular_function and ('directly positively regulates' some 'catalytic activity')
regulation of gene expression
  • GP-A [regulation of gene expression Z] has_regulation_target GP-B
  • These can be modeled as GP-A<-enabled_by-[root MF]-part_of->[regulation of Z]-has_input->GP-B,-causally upstream of (positive/negative effect)->[root MF]-enabled_by->GP-B
  • Will be positively or negatively causally upstream of instead, if in the original term.
regulation of protein metabolic process
  • GP-A [regulation of protein metabolic process Z, not including cases above, or regulation of protein localization] has_regulation_target GP-B
  • These are GO:0019538 and the descendants that are not also descendants of terms listed above
  • These can be modeled as GP-A<-enabled_by-[root MF]-part_of->[regulation of Z]-has_input->GP-B
regulation of protein biosynthetic process
  • GP-A [regulation of protein biosynthetic process Z] has_regulation_target GP-B
  • These can be modeled as GP-A<-enabled_by-[root MF]-part_of->[regulation of Z]-has_input->GP-B,-causally upstream of (positive/negative effect)->[root MF]-enabled_by->GP-B
  • Will be positively or negatively causally upstream of instead, if in the original term.
regulation of protein catabolic process
  • GP-A [regulation of protein catabolic process Z] has_regulation_target GP-B
  • GO:0042176 and descendants
regulation of receptor internalization
  • GP-A [Regulation of receptor internalization (GO:0002092) Z] has_regulation_target GP-B
  • These are GO:0002092 and descendants
  • These can be modeled as GP-A<-enabled_by-[root MF]-part_of->[regulation of Z]-has_input->GP-B,-causally upstream of (positive/negative effect)->[root MF]-enabled_by->GP-B
  • If term is positive regulation, relation will be causally upstream of, negative effect; if negative reg, relation will be causally upstream of, positive effect
regulation of protein localization
  • GP-A [regulation of protein localization] has_regulation_target GP-B
  • GO:0032880 and descendants
  • These can be modeled as GP-A<-enabled_by-[root MF]-part_of->[regulation of Z]-has_input->GP-B

Multiple Extensions

Rules for importing annotations with multiple, comma-separated annotation extensions.

Chained Relations

  • regulates_o_results_in_specification_of
  • regulates_o_results_in_movement_of

GPAD Output Files