Tornike Onoprishvili

The Tox21 challenges data scientists around the world to predict chemical interactions with biological pathways. I developed a statistical model—ensemble (RF, GBM, and DT) and tuned the model via Bayesian optimization to address a specific sub-task in this challenge. With 50+ teams competing to build the best predictor, my model was second due to my hand-crafted innovative feature engineering techniques.

In this work, we present a GPU-accelerated cosine similarity implementation for tandem mass spectrometry (MS), with an approximately 1000-fold speedup compared to the MatchMS reference implementation, without any loss of accuracy. This improvement enables repository-scale spectral library matching for compound identification without the need for large compute clusters. This impact extends to any spectral comparison-based methods, such as molecular networking approaches and analogue search.