Helixer–ab initio Prediction of Primary Eukaryotic Gene Models Combining Deep Learning and a Hidden Markov Model. Trained models.

Models for structural gene annotation of vertebrates, invertebrates, plants and fungi, to be used with Helixer (https://github.com/weberlab-hhu/Helixer). In summary, input is one hot-encoded DNA sequence and output is basewise-predictions, which are automatically fed into an HMM to produce finalized primary gene models. This upload includes both during development and final 'best models' as described here: https://www.nature.com/articles/s41592-025-02939-1

While model details, ranking and meta info can be found here: https://github.com/weberlab-hhu/Helixer/blob/main/resources/model_list.csv, we recommend simply using the autodownload. 

[this version adds a mammal model (mammal_v0.3_m_0100.h5)]