Back

Natural language processing charts transcriptome evolution to design combination cancer therapies

Jassim, A.; Nimmervoll, B. V.; Terranova, S.; Nathan, E.; Masih, K. E.; Ruff, L.; Duarte, M.; Cooper, E.; Hu, L. P.; Katyal, G.; Akhbari, M.; Gilbertson, R. J.; Terhune, C.; Balmus, G.; Jackson, S. P.; Filbin, M. G.; Hill, A.; Patrizi, A.; Dani, N.; Regev, A.; Lehtinen, M. K.; Gilbertson, R. J.

2024-04-09 cancer biology
10.1101/2024.04.09.585681 bioRxiv
Show abstract

Combination treatment-the mainstay of cancer therapy-often fails because treatment interactions evoke complex resistance mechanisms that are hard to predict. Designing combination therapy to prevent treatment resistance is especially challenging for rare cancers. Here, we introduce REsistance through COntext DRift (RECODR): a computational pipeline that tracks how genes change their transcriptome context across cancer development and drug treatment conditions. By applying RECODR to a genetically modified mouse model of choroid plexus carcinoma-a rare brain tumour of young children-we identified patterns of transcriptome evolution, cellular heterogeneity and treatment targets that emerged as tumours were initiated and resisted combination treatment. This enabled the prediction of treatment resistance mechanisms and the design of highly effective therapeutic protocols that avoided treatment failure. RECODR can describe complex and dynamic changes in normal and diseased tissues that could be applied to multimodal data from a variety of settings, mitigating treatment resistance across cancers and other diseases.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.