Track type / EDAM ontology collaboration

[sveinugu]

• [x] Go through EDAM and locate useful track types, and suggest others that can be added
• [ ] Contact EDAM with suggestions
• [x] In the meantime, use list as a limited vocabulary in the schema so that we do not depend on EDAM implementations.

[sveinugu]

@dzerbino @jmfernandez @ehovig After much consideration, i suggest the property should be named “type_of_condensed_data». Better would have been «condensed_data_type», but it is grammatically wrong I think, as it seems to say that it is the type that is condensed and not the data.

“Data condensation» is a term that has replaced «data reduction» because of the negative connotations of the latter term. All tracks contain condensed data by their very nature. But what is lacking everywhere is a vocabulary which describes the generic form or type of the condensed data (as a “track subtype”), including how it should be interpreted. The idea is that these terms are purely technical and not biological. Here are some suggested terms:

• Sequence-derived regions (Regions derived primarily from reference sequence, including the reference genome)
• Experimentally-derived regions (Regions derived primarily from a single or a few experiments)
• Predicted regions (Regions derived from reference sequences and/or aggregated experimental data, where the act of prediction is central)
• Predicted segmentation (Partition of the reference genome into predicted segments derived from reference sequences and/or aggregated experimental data, where the act of prediction is central)
• Population-derived variants (Sequence variation from the reference genome, derived from larger collections of experimental data from a population sample)
• Individual variants - (Sequence variation from the reference genome, derived from experimental data of a single individual)
• Peaks (Predicted peaks of high read coverage)
• Broad peaks (Predicted broad peak regions of high read coverage)
• Narrow peaks (Predicted narrow peak regions of high read coverage)
• Gapped peaks (Predicted gapped peak regions of high read coverage)
• Signal values (Signal-to-noise value per base pair or homogeneous region)
• Read coverage (Coverage of mapped reads per base pair or homogeneous region)
• Read counts (Counts of mapped reads per base pair or homogeneous region)
• Mapped single-end reads (Single-end sequenced reads mapped to a reference genome)
• Mapped paired-end reads (Paired-end sequenced reads mapped to a reference genome)

[sveinugu]

In addition, I suggest to add a “track_type” property using the terminology of our 15 track types: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-12-494/figures/1

Of course I am partial here, but there exists no other terminology of the kind, and they have proved to be very useful.