Some 'Diseases' classified as 'Side Effects'

[mzaleski1]

Hi Daniel,

I am interested in using het.io as a tool to decide which small molecule drugs (and combinations of those drugs) could be promising treatments for acute illnesses.

I've tried using the Node Search function to see which compounds are connected with these diseases. Specifically, I searched for ARDS (Acute Respiratory Distress Syndrome), Ischaemic stroke, and Sepsis, but they only showed up as a side effects, not diseases.

Am I doing something wrong with the search, or is this just the way the database is curated?
Forgive me if this is an obvious question, I'm new to network pharmacology and these types of databases!

Thanks,
Mike

[dhimmel]

Thanks Mike for your question.

I searched for ARDS (Acute Respiratory Distress Syndrome), Ischaemic stroke, and Sepsis, but they only showed up as a side effects, not diseases.

Unfortunately, those diseases were not included in Hetionet v1.0 as diseases. While they do show up as side effects, the only relationship we have for side effects is Compound-causes-Side Effect. Hence, those Side Effects nodes won't be very helpful for understanding a disease.

Am I doing something wrong with the search, or is this just the way the database is curated?

You are doing things right. I'm copying a section of the Rephetio Manuscript describing how we decided what disease nodes to include:

We selected 137 terms from the Disease Ontology [77,78] (which we refer to as DO Slim [79,80]) as our disease set. Our goal was to identify complex diseases that are distinct and specific enough to be clinically relevant yet general enough to be well annotated. To this end, we included diseases that have been studied by GWAS and cancer types from TopNodes_DOcancerslim [81]. We ensured that no DO Slim disease was a subtype of another DO Slim disease.

Expand for Bibliography

77. Disease Ontology: a backbone for disease semantic integration
L. M. Schriml, C. Arze, S. Nadendla, Y.-W. W. Chang, M. Mazaitis, V. Felix, G. Feng, W. A. Kibbe
Nucleic Acids Research (2011-11-12) https://doi.org/fwvx82
DOI: 10.1093/nar/gkr972 - PMID: 22080554 - PMCID: PMC3245088

78. Disease Ontology 2015 update: an expanded and updated database of human diseases for linking biomedical knowledge through disease data
Warren A. Kibbe, Cesar Arze, Victor Felix, Elvira Mitraka, Evan Bolton, Gang Fu, Christopher J. Mungall, Janos X. Binder, James Malone, Drashtti Vasant, ... Lynn M. Schriml
Nucleic Acids Research (2014-10-27) https://doi.org/f3mn6b
DOI: 10.1093/nar/gku1011 - PMID: 25348409 - PMCID: PMC4383880

79. Unifying disease vocabularies
Daniel Himmelstein, Tong Shu Li
ThinkLab (2015-03-30) https://doi.org/f3mqv5
DOI: 10.15363/thinklab.d44

80. User-Friendly Extensions To The Disease Ontology V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/f3mqvc
DOI: 10.5281/zenodo.45584

81. Generating a focused view of disease ontology cancer terms for pan-cancer data integration and analysis
T.-J. Wu, L. M. Schriml, Q.-R. Chen, M. Colbert, D. J. Crichton, R. Finney, Y. Hu, W. A. Kibbe, H. Kincaid, D. Meerzaman, ... R. Mazumder
Database (2015-04-04) https://doi.org/f3mn6c
DOI: 10.1093/database/bav032 - PMID: 25841438 - PMCID: PMC4385274

When building Hetionet v1.0, our focus was complex diseases with sufficient information to be well-connected. Perhaps today, we could rebuild the network with more diseases. But until then, you won't be able to directly infer things about "ARDS (Acute Respiratory Distress Syndrome), Ischaemic stroke, and Sepsis".

An alternative option if you're an expert in those conditions would be to start with genes, pathways, or symptoms related to those diseases and see if you get anwhere.

[mzaleski1]

Thanks, Daniel!
I will try searching genes, pathways, and symptoms and see how it goes.

I imagine it would be a significant amount of work to rebuild the network with more diseases? I will see if people in my lab are interested in helping with this kind of work. Maybe we could discuss in more detail in the next couple weeks? Since we can't go into lab right now, this kind of computational/database project could be perfect for us now.

[dhimmel]

I imagine it would be a significant amount of work to rebuild the network with more diseases? I will see if people in my lab are interested in helping with this kind of work.

Yes it's a lot of work since many of the 29 input datasets would have to be reprocessed with the new disease set. The place where all the datasets are combined is here.

Since we can't go into lab right now, this kind of computational/database project could be perfect for us now.

Happy to help anyone in your lab get going in the right direction. Ideally, the person would have experience with Git, Python, Pandas, and Jupyter notebooks (or be willing to spend some extra time learning).