Recursion recently published a paper in Nature Biotechnology on TxPert, a new method that can predict how cells respond to genetic changes — even in conditions they haven’t seen before. Its work in the space has continued to evolve, with a state-of-the-art transcriptomic model, TxFM.
Clinical stage TechBio company Recursion just published a paper in Nature Biotechnology on TxPert, a new method that can predict how cells respond to genetic changes — even in conditions they haven’t seen before. Since that paper was accepted, Recursion’s work in modelling transcriptomics has evolved considerably.
At the recent ICLR conference (and upcoming at ICML), the company presented TxFM, a new state-of-the-art transcriptomics foundation model that takes a radically different approach to “bigger is better” models. In fact, via rigorous benchmarking, the paper demonstrates that massive foundation models fail to outperform simple statistical methods.
Recursion’s model combines its proprietary dataset with a highly customized Masked Autoencoder (MAE) architecture to represent genes as they appear in nature (i.e., unordered). It provides a much more accurate representation of how genes interact as a system, connects experimental results directly to patient biology, and outperforms models up to 100x larger in terms of data size.
