BMFM-RNA: whole-cell expression decoding improves transcriptomic foundation models

Transcriptomic foundation models pretrained with masked language modeling can achieve low pretraining loss yet produce poor cell representations for downstream tasks. We introduce whole-cell expression decoding (WCED), where models reconstruct the entire gene vocabulary from a single CLS token embedding, even with limited inputs, creating a maximally informative bottleneck. WCED consistently outperforms MLM on all downstream metrics despite higher reconstruction error during training. Gene-level error tracking reveals that both methods preferentially learn genes whose expression co-varies with stable transcriptional programs rather than those driven by transient factors. We further add hierarchical cross-entropy loss that exploits Cell Ontology structure for zero-shot annotation at multiple granularity levels. Models trained with these objectives achieve best overall performance across CZI benchmarks, on zero-shot batch integration and linear probing cell-type annotation. Methods are implemented in biomed-multi-omic ( https://github.com/BiomedSciAI/biomed-multi-omic ), an open-source framework for transcriptomic foundation model development.