Mapping Walker

This application will:

• crawl a mapping endpoint weaving a set of SSSOM mappings
• input these into boomer
• visualize the results with obographviz

THIS IS ALPHA SOFTWARE

Requirements

You need the following on your path:

• graphviz
• boomer
• og2dot

On MacOS you may need to install openblas via homebrew and set the OPENBLAS environment variable before installing dependencies:

brew install openblas
export OPENBLAS="$(brew --prefix openblas)"
poetry install

Running

mapcrawl OMIM:603467

you can enter as many CURIEs as you like

The output looks like:

Note the visualization doesn't show the mappings that were rejected - see the full boomer report file. It shows things like:

• Fanconi anemia SiblingOf OMIM_614087 0.09999999999999998
• Fanconi anemia ProperSuperClassOf OMIM_614087 (most probable) 0.8
• aplastic anemia EquivalentTo Anemia, Aplastic (most probable) 0.8
• Fanconi anemia EquivalentTo Fanconi Anemia (most probable) 0.8

Gallery

Anatomy

• FMA:24879

these are the full results:

• Lower Extremity EquivalentTo Lower Extremity (most probable) 0.8
• hindlimb EquivalentTo Hindlimb (most probable) 0.8
• hindlimb EquivalentTo hindlimb (most probable) 0.8
• hindlimb EquivalentTo hindlimb (most probable) 0.8
• hindlimb SiblingOf 0.09999999999999998
• hindlimb EquivalentTo Hindlimb (most probable) 0.8
• Lower Extremity SiblingOf 0.09999999999999998
• hindlimb EquivalentTo (most probable) 0.8
• hindlimb SiblingOf Lower Extremity 0.09999999999999998
• hindlimb EquivalentTo hindlimb (most probable) 0.8
• hindlimb EquivalentTo obsolete_hindlimb (most probable) 0.8
• hindlimb EquivalentTo hindlimb (most probable) 0.8
• hindlimb EquivalentTo Free lower limb (most probable) 0.8
• Hindlimb EquivalentTo Hindlimb (most probable) 0.8
• hindlimb EquivalentTo lower limb (most probable) 0.8
• leg EquivalentTo Lower Extremity (most probable) 0.8
• Lower Extremity EquivalentTo (most probable) 0.8
• hindlimb EquivalentTo hindlimb (most probable) 0.8

note boomer effectively rejected some links (SiblingOf) that came from some mappings

Disease

MONDO:0019249 ! mucopolysaccharidosis

How it works

It currently queries OxO - it will do a breadth first search until either saturated or either max depth or max queries is reached

All configuration is specified in LinkML yaml - docs will be up soon

Currently it applies the standard SSSOM-py mappings for all predicates obtained from OxO (most mappings in OxO are RELATED but there are some narrow/broad)

Default confidence is 0.8 (currently no mappings in OxO have confidence)

We don't yet allow plugin in of rules - e.g. boost confidence if mapping comes from a more trusted source, or is HumanCurated

TODO

• fix CURIEs
• add bioportal endpoint (currently uses oxo)
• add ubergraph endpoint