Selene

Selene enables development, training, and application of PyTorch-based deep learning models for biological sequence data, including DNA sequences, to study genomic function and variation.

Key Features:

• PyTorch foundation: Built on PyTorch to implement and train deep learning architectures.
• Model architecture support: Supports training of published model architectures and development of custom deep learning models.
• Sequence data support: Applicable to various types of biological sequence data, including DNA sequences.

Scientific Applications:

• Genomic function prediction: Enables modeling of sequence-to-function relationships to advance understanding of genomic functions.
• Variant interpretation: Supports analysis to identify genetic variations linked to disease.
• Therapeutic research: Facilitates model-based investigation relevant to developing new therapeutic strategies.

Methodology:

Model training using existing or custom deep learning architectures; evaluation of model performance to enable iterative improvements; application of trained models to biological datasets to derive insights.