Should subtypes of Bardet-Biedl have the same treatments

[cmungall]

My understanding is that knowing which gene is responsible for BBL does not drastically affect treatment

There are some statistical variations reported in https://pubmed.ncbi.nlm.nih.gov/20301537/ https://www.ncbi.nlm.nih.gov/books/NBK1363/#bbs.Causes_of_BardetBiedl_Syndrome

And from my naive non-clinical viewpoint I imagine that this might be reflected in statistical variations in treatments - e.g. if obesity is more severe in one form we might expect to see dietary interventions employed more. But do we want to curate at this level, or just lump all BBS together?

It looks like when we annotate this NBK we make semi-redundant annotations to all subtypes; this is the distibution of annotations by disease:

✗ shep -d maxoa -c maxoa fq -w "citation: PMID:20301537" -S disease_name
{
  "disease_name": {
    "Bardet-biedl Syndrome 1": 4,
    "Bardet-biedl Syndrome 2": 2,
    "Bardet-biedl Syndrome 21": 2,
    "Bardet-biedl Syndrome 9": 2,
    "Bardet-biedl Syndrome 3": 1,
    "Bardet-biedl Syndrome 4": 1,
    "Bardet-biedl Syndrome 5": 1,
    "Bardet-biedl Syndrome 7": 1,
    "Bardet-biedl Syndrome 11": 1,
    "Bardet-biedl Syndrome 16": 1,
    "Bardet-biedl Syndrome 19": 1,
    "Bardet-biedl Syndrome 20": 1,
    "Bardet-biedl Syndrome 14": 1,
    "Bardet-biedl Syndrome 6": 1,
    "Bardet-biedl Syndrome 10": 1,
    "Bardet-biedl Syndrome 18": 1,
    "Bardet-biedl Syndrome 8": 1,
    "Bardet-biedl Syndrome 12": 1,
    "Bardet-biedl Syndrome 13": 1
  }
}

We can look at BBS2 ("leanest of obesity phenotype"), this has dietary interventions for obesity and diabetes:

[
  {
    "disease_id": "MONDO:0014432",
    "disease_name": "Bardet-biedl Syndrome 2",
    "citation": "PMID:20301537",
    "maxo_id": "MAXO:0000088",
    "maxo_label": "dietary intervention",
    "hpo_id": "HP:0001513",
    "maxo_relation": "TREATS",
    "evidence_code": "TAS",
    "extension_id": "",
    "extension_label": "",
    "attribute": "",
    "creator": "ORCID:0000-0001-7941-2961",
    "last_update": "2023-01-31",
    "created_on": "2022-02-15"
  },
  {
    "disease_id": "MONDO:0014432",
    "disease_name": "Bardet-biedl Syndrome 2",
    "citation": "PMID:20301537",
    "maxo_id": "MAXO:0000088",
    "maxo_label": "dietary intervention",
    "hpo_id": "HP:0000819",
    "maxo_relation": "TREATS",
    "evidence_code": "TAS",
    "extension_id": "",
    "extension_label": "",
    "attribute": "",
    "creator": "ORCID:0000-0001-7941-2961",
    "last_update": "2023-01-31",
    "created_on": "2022-02-15"
  }
]

and BBS4 ("Early-onset morbid obesity"); this only has dietary interventions for obesity:

[
  {
    "disease_id": "MONDO:0014433",
    "disease_name": "Bardet-biedl Syndrome 4",
    "citation": "PMID:20301537",
    "maxo_id": "MAXO:0000088",
    "maxo_label": "dietary intervention",
    "hpo_id": "HP:0001513",
    "maxo_relation": "TREATS",
    "evidence_code": "TAS",
    "extension_id": "",
    "extension_label": "",
    "attribute": "",
    "creator": "ORCID:0000-0001-7941-2961",
    "last_update": "2023-02-07",
    "created_on": "2022-02-15"
  }
]

Elsewhere things seem to line up with existing HPO annotations. E.g. physical therapy is a treatment for gait ataxia, in BBS1 and not the other forms, which matches HPO annotations. But is it really the case that this phenotype is restricted to BBS1 or is there just a lot of variable penetrance? If we annotate a gait phenotype to another BBS do we then update the maxo file?

It seems more scalable to just annotate to the generic Mondo BBS, and infer phenotype-specific treatments

[pnrobinson]

@cmungall The literature has a mix -- the above are for specific forms of BBS, but there are papers that are for BBS in general, for ciliopathies in general etc. I think we should curate to reflect the assertions made in the literature and then leverage the structure of MONDO to infer that a BBS4 annotation might have some weight for BBS5 etc. While the BBS case is probably straightforward, many are not and if we curate what the papers assert then we can adjust what the websites show etc. using up to date knowledge. Thoughts?