Kyle Swanson

RL-Finetuning of OpenAI o1-mini to Enhance Biomedical Reasoning

This study evaluates the effectiveness of RL finetuning o1-mini for biomedical reasoning tasks and finds mixed results in gene classification and small molecule prediction.

SyntheMol-RL: a flexible reinforcement learning framework for designing novel and synthesizable antibiotics

SyntheMol-RL generates novel and synthesizable antibiotics by optimizing for antibacterial activity and solubility, outperforming prior models and showing in vitro and in vivo efficacy.

The Virtual Lab: AI Agents Design New SARS-CoV-2 Nanobodies with Experimental Validation

Role-specialised LLM agents under an LLM principal investigator, with light human steering, assembled an ESM/AlphaFold-Multimer/Rosetta pipeline and designed 92 SARS-CoV-2 nanobodies; two bound recent variants better.

ADMET-AI: A machine learning ADMET platform for evaluation of large-scale chemical libraries

ADMET-AI, a Chemprop-RDKit graph neural network trained on 41 TDC datasets, has the best average leaderboard rank and scores one million molecules in 3.1 hours.

Towards Universal Cell Embeddings: Integrating Single-cell RNA-seq Datasets across Species with SATURN

SATURN integrates single-cell RNA-seq datasets across species using deep learning to create universal cell embeddings, enabling cross-species analysis and annotation transfer.

7-UP: generating in silico CODEX from a small set of immunofluorescence markers

7-UP generates in silico 40-plex CODEX data from a 7-plex mIF panel using cellular morphology, enabling cell type classification and survival prediction across diverse samples.

Predicting Immune Escape with Pretrained Protein Language Model Embeddings

The paper predicts immune escape by using pretrained protein language model embeddings, showing they outperform zero-shot methods but are not superior to simple baselines.

SPACE-GM: geometric deep learning of disease-associated microenvironments from multiplex spatial protein profiles.

SPACE-GM identifies spatial motifs in cancer samples linked to recurrence and immunotherapy outcomes using geometric deep learning.