ML prediction for DAC combination therapies

[abearab]

• Build knowledge graphs
• Define ML tasks
  Analysis:
• https://github.com/GilbertLabUCSF/Decitabine-treatment/blob/main/DAC_combo.ipynb

[abearab]

Related tools:

• https://github.com/mims-harvard/TxGNN
• https://github.com/mims-harvard/TDC/blob/master/tutorials/TDC_104_ML_Model_DeepPurpose.ipynb

---

Igor F Tsigelny, Artificial intelligence in drug combination therapy, Briefings in Bioinformatics, 2019, https://doi.org/10.1093/bib/bby004

Jafari, M., Mirzaie, M., Bao, J. et al. Bipartite network models to design combination therapies in acute myeloid leukaemia. Nat Commun 13, 2128 (2022). https://doi.org/10.1038/s41467-022-29793-5

[abearab]

40988f9

[abearab]

I think the evidence I found regarding current combinations of DNMT + DHODH inhibitors can be a proof of concept for this drug treatment model using multi-omics readouts. Thus, it might be useful to think about it for defining new tasks.

---

Related to mims-harvard/TDC#207 and mims-harvard/PrimeKG#12

[abearab]

https://tdcommons.ai/benchmark/drugcombo_group/overview/