Multi-species code

[fbastian]

In GitLab by @fbastian on Apr 19, 2016, 16:06

Rough figure of workflow for multi-species analyses in bgee-core:

multispe_workflow.pdf

The Stream zipping, and call accumulation steps, are not needed for now, as we only need to compute conservation score for single gene page.

Description of the workflow for single gene page:

MAPPING (assigned to @pmoret)

• create method generating Map<OMANodeId, Set<Gene>> to get orthology info (arguments: one taxon ID, Collection of targeted species IDs)

• create class to manage organ homology (with CIO support, reference lists, ,…). Maybe named 'AnatEntitySimilarity'?

• create class to manage dev. stage similarity (with CIO support, reference lists, ,…). Maybe named 'DevStageSimilarity'?

• create class MuliSpeciesCall. Will store info about organ homology groups, single-species call, maybe gene orthology, …

• create method returning Collection of AnatEntitySimilarity for a specific taxon (argument: one Taxon ID, collection of targeted species IDs)

• create method returning Collection of DevStageSimilarity for a specific taxon (argument: one Taxon ID, collection of targeted species IDs)

PROPAGATION (assigned to @vrechdelaval):

• create method to propagate ExpressionCalls (can take as arguments lists of targeted organs/stages)
• create method to propagate NoExpressionCalls (can take as arguments lists of targeted organs) (lower priority maybe)

GROUPING (in case of single gene page, grouping performed once per taxon)

• method to reconcile calls for a single-gene: either for a single organ, or for a group of homologous organs, reconcile calls (e.g., expr affy vs. no-expr RNA-Seq). The point is that this method can accept arbitrary number of calls to perform reconciliation, it doesn't matter whether calls come from homologous organs or from single species Return one representative call (with reconciled quality per data types, observed data state, conflict status etc. But not with organId-stageId) => assigned to @vrechdelaval

• main method arguments: Set<ExpressionCallTO>, Set<NoExpressionCallTO>, Set<DiffExpressionCallTO> (already propagated as needed). All these calls are for one single gene orthology group at a specific taxonomic level

• main method argument: Collection of HomologousOrganGroup => 2 signatures, one returning Calls, one returning MultiSpeciesCalls

[fbastian]

In GitLab by @vrechdelaval on May 17, 2016, 18:18

• Create MultiSpeciesOntology (which extends Ontology) using TaxonConstraintTO, and add methods retrieving entities and relations according provided species.
• Update ConditionUtils to accept MultiSpeciesOntology

=> assigned to @vrechdelaval

[fbastian]

In GitLab by @pmoret on May 19, 2016, 13:09

[fbastian]

In GitLab by @fbastian on May 20, 2016, 11:50

I wonder who draw this shitty figure...