Las redes neuronales entrenadas con texto genético saben 'leer' el ADN bacteriano. Los científicos tomaron el potente modelo Evo 2 y le añadieron un minúsculo sensor ajustable — una sonda genética, sin tocar el modelo en sí. La sonda capta señales relacionadas con genes peligrosos, como los de resistencia a antibióticos. De forma similar a la espectroscopía — un método para analizar sustancias por su brillo característico — la sonda reconoce la 'respuesta' específica del modelo al código genético.
Este enfoque resultó sorprendentemente preciso. Incluso una sonda simple ofrece buenos resultados, y con un mecanismo de atención (attention) la precisión alcanza casi 0,99: distingue subtipos de resistencia a fármacos. Y, como un detector de metales, funciona con fragmentos de ADN. Esto permite analizar de forma barata y rápida las aguas residuales hospitalarias, detectando amenazas ocultas en las alcantarillas y alertando sobre brotes infecciosos antes de que aparezcan síntomas en las personas.
tal como lo escribieron los autores, en inglés arXiv:2607.14070
Genomic foundation models such as Evo 2 learn rich sequence representations, but their value for biosecurity screening is largely unexplored. We ask how much biosecurity-relevant signal is linearly accessible in these representations by training minimal linear and attention probes on frozen Evo 2 layer-26 activations, without fine-tuning the underlying model. Across held-out metagenomic test sets, the probes detect antimicrobial resistance (AMR) with strong discrimination: a linear probe reaches a region-level ROC-AUC of 0.888 (mean-pool), rising to 0.977 with a single-head attention probe. The probes resolve finer-grained AMR drug-class subcategories and separate them from unrelated functional genes, providing additional evidence that the learned signal is not explained solely by generic functional-gene status. Bacterial virulence is also decodable, though more weakly (region-level ROC-AUC 0.833). The AMR probe retains comparable ranking performance on simulated short reads without retraining, enabling evaluation before assembly in settings where assembly is computationally costly or unreliable. It achieves a read-level ROC-AUC of 0.898 (mean-pool), comparable to the mean-pooled full-region result. Within SynGenome, AMR-associated prompt labels are only weakly recoverable from Evo 1.5-generated sequences; these prompt-derived labels do not establish the function of the generated response sequences. A complementary sparse-autoencoder analysis recovers interpretable resistance-associated features but proves less consistent than the supervised probes. Together, these results position lightweight embedding-based probes as a fast, inexpensive first-pass detection layer for metagenomic biosurveillance and map both strengths and current limits of the approach. This work was conducted as part of the AIxBio Hackathon 2026 hosted by BlueDot Impact, Apart Research, and Cambridge Biosecurity Hub.
La resistencia a los antibióticos es una de las principales amenazas para la salud: según las previsiones, para 2050 podría causar 10 millones de muertes al año.