Back

HAVEN: Hierarchical Attention for Viral protEin-based host iNference

Antony, B.; Haghani, M.; Lauring, A. S.; Karpatne, A.; Murali, T. M.

2025-06-13 bioinformatics
10.1101/2025.06.09.658367 bioRxiv
Show abstract

It is crucial to accurately predict hosts of viruses to understand and anticipate human infectious diseases that originate from animals. There is a lack of versatile models that handle out-of-distribution factors such as unseen hosts and viruses. We develop a machine learning model for predicting the host infected by a virus, given only the sequence of a protein encoded by the genome of that virus. Our approach, HAVEN, is the first to apply to multiple hosts and to generalize to unseen hosts and viruses. HAVEN is a transformer-based architecture coupled with hierarchical self-attention that can accept sequences of highly diverse lengths. We integrate HAVEN with a prototype-based few-shot learning classifier to predict rare classes. We demonstrate the accuracy, robustness, and generalizability of HAVEN through a comprehensive series of experiments. In particular, we show that HAVEN can achieve a median AUPRC of 0.67 while predicting common hosts. Moreover, HAVEN retains this AUPRC value even for rare hosts (median prevalence as low as 0.09%). Our model performs on par with state-of-the-art foundation models, which are 65 to 5, 000 times larger in size, and outperforms them in identifying hosts of SARS-CoV-2 variants of concern.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.