Bugfix - Showing unappropriate organ anatomogram view

[lingyun1010]

@silvie menthioned that something went wrong with this experiment - it doesn't have inferred cell type information, so anatomogram shouldn't be displayed on the experiment page but it is there (empty): https://wwwdev.ebi.ac.uk/gxa/sc/experiments/E-CURD-126/results/tsne.

[sfexova]

Hi, so I think the rules should be as follows:

• Human anatomograms

1. species is "Homo sapiens"
2. [organism part] is or maps up to an organ for which we have an anatomogram
3. DOES NOT have Characteristics[cell line]
4. [developmental stage] is NOT embryo or fetus (and their synonyms or children terms in the ontology) and doesn't contain either of the terms and their derivatives (embryonic, foetus, fetal, foetal) even in a string
5. the experiment has authors cell types and at least one of these cell type annotations maps to an annotation in the relevant anatomogram

• Fly anatomograms

1. species is "Drosophila" sth (99% will be Drosophila melanogaster but potentially can be other Drosophila species ananassae, virilis, etc etc)
2. [organism part] is or maps up to an organ for which we have an anatomogram
3. DOES NOT have Characteristics[cell line]
4. [developmental stage] IS adult (or synonym/children terms in the ontology)
5. the experiment has authors cell types and at least one of these cell type annotations maps to an annotation in the relevant anatomogram

[sfexova]

I think the above rules should filter out vast majority of unwanted display cases but it would still be good to implement a way to specifically 'ban' certain experiments where we do not want the anatomogram shown even if they pass the filtering criteria

[lingyun1010]

Hi, so I think the rules should be as follows:

* Human anatomograms


1. species is "Homo sapiens"

2. [organism part] is or maps up to an organ for which we have an anatomogram

3. DOES NOT have Characteristics[cell line]

4. [developmental stage] is NOT embryo or fetus (and their synonyms or children terms in the ontology) and doesn't contain either of the terms and their derivatives (embryonic, foetus, fetal, foetal) even in a string

5. the experiment has authors cell types and at least one of these cell type annotations maps to an annotation in the relevant anatomogram


* Fly anatomograms


1. species is "Drosophila" sth (99% will be Drosophila melanogaster but potentially can be other Drosophila species ananassae, virilis, etc etc)

2. [organism part] is or maps up to an organ for which we have an anatomogram

3. DOES NOT have Characteristics[cell line]

4. [developmental stage] IS adult (or synonym/children terms in the ontology)

5. the experiment has authors cell types and at least one of these cell type annotations maps to an annotation in the relevant anatomogram

Hi @sfexova , regarding the rules above, I have a few questions for example human anatomogram:

1. If we need to check sdrf file or condesed-sdrf file for those fields, including [organism part], Characteristics[cell line], [developmental stage], are they mapping to any field in solr?
2. Can you explain more for Rule 4, espcially

• whose synonyms or children terms in the ontology and how to check those terms?
• what doesn't contain either of which terms and whose derivatives?

3. Regarding Rule 5, could you give an example in one or two experiments?

Thank you!

[ke4]

I think the above rules should filter out vast majority of unwanted display cases but it would still be good to implement a way to specifically 'ban' certain experiments where we do not want the anatomogram shown even if they pass the filtering criteria

If we need some sort of black-list then I would not hardcode it to any class variable, but rather I would put them into a properties file, that we could modify (add/remove accessions) without any code modification. That way we can add/remove accessions without recompile the app.
The relevant class could read that properties file and use those values (accessions) to ban experiments.

[sfexova]

Hi sorry, slipped from my radar:

• If we need to check sdrf file or condesed-sdrf file for those fields, including [organism part], Characteristics[cell line], [developmental stage], are they mapping to any field in solr?

  Don't know about the mapping in Solr, please check with the dataprods

Can you explain more for Rule 4, espcially

• whose synonyms or children terms in the ontology and how to check those terms?
  

  not sure about the implementation but the idea is to check that the [developmental stage] in the human experiments is NOT annotated with the value 'embryo stage' (http://www.ebi.ac.uk/efo/EFO_0007725) or 'embryo' (http://purl.obolibrary.org/obo/UBERON_0000922) + their synonyms from the ontology and/or doesn't contain a value (even unmapped to any ontology term) that includes the string 'embryo'
  the same for 'fetus' or 'foetus' (different spellings, same thing): [developmental stage] value doesn't contain ontology terms 'fetal stage' http://purl.obolibrary.org/obo/HsapDv_0000037) and 'fetus' (http://purl.obolibrary.org/obo/BTO_0000449) + their recognised synonyms in the ontology and/or doesn't contain a values (even unmapped) that includes 'fetal', 'foetal', 'fetus' or 'foetus'

• what doesn't contain either of which terms and whose derivatives?
  

  see above

• Regarding Rule 5, could you give an example in one or two experiments?

  e.g. E-GEOD-130148 (lung), E-MTAB-5061 (pancreas), E-MTAB-6701 (placenta)