Search papers, labs, and topics across Lattice.
GeneICL, a 4.2M-parameter tabular foundation model, addresses the challenges of high dimensionality and limited labeled data in bulk transcriptomics by employing a semi-synthetic pretraining prior derived from measured expression profiles and a recurrent architecture. This approach enables training-free reduction to regression for right-censored survival prediction, outperforming both self-supervised transcriptomic models and tuned baselines across 80 clinical outcome-prediction tasks. GeneICL achieves state-of-the-art performance with significantly fewer parameters and rapid inference, demonstrating the efficacy of transcriptomics-aware pretraining over sheer scale.
Transcriptomics-specific pretraining, not model scale, unlocks state-of-the-art clinical outcome prediction from bulk RNA-seq, even for survival tasks, with a 4.2M parameter model.
Gene expression is widely measured in biomedicine, yet clinical outcome prediction remains challenging due to high dimensionality, strong feature correlations, and limited labeled data. Large self-supervised transcriptomic foundation models often fail to outperform simple supervised baselines. Tabular foundation models offer an alternative through in-context learning, but are typically pretrained on generic synthetic data rather than transcriptomic structure. We ask whether transcriptomics-aware pretraining, rather than scale, is the missing ingredient. Towards this end, we introduce GeneICL, a 4.2M-parameter tabular foundation model combining a semi-synthetic pretraining prior built from measured bulk expression profiles with a parameter-efficient recurrent architecture. We further enable right-censored survival prediction via a training-free reduction to regression using Cox partial-likelihood residuals. We evaluate GeneICL on 80 clinical outcome-prediction tasks spanning classification, regression, and survival. Tabular foundation models consistently outperform self-supervised transcriptomic models, while GeneICL achieves the best overall rank among evaluated foundation models and tuned baselines. GeneICL does so with up to 387$\times$ fewer parameters, no gradient updates at inference, and predictions within seconds on a laptop CPU.
