autoBioSeqpy

autoBioSeqpy applies deep learning to classify biological sequences for tasks including prediction of Type III secreted proteins, protein subcellular localization, and CRISPR/Cas9 sgRNA activity.

Key Features:

• Automated Workflow: Automates text reading, parameter initialization, sequence encoding, model loading, training, and evaluation.
• Customizable Steps: Allows customization of modeling steps and workflows.
• Model Templates: Includes ready-to-use and adaptable deep learning model templates.
• Sequence Encoding: Implements sequence encoding methods for biological sequences.
• Parameter Initialization: Provides parameter initialization for models.
• Model Loading: Supports loading of pretrained or user-specified models.
• Training: Supports training of deep learning architectures on sequence data.
• Evaluation: Provides evaluation procedures for trained models.
• Deep Learning Frameworks: Implements architectures using TensorFlow, PyTorch, and Keras.

Scientific Applications:

• Prediction of Type III Secreted Proteins: Identifies proteins secreted via type III secretion systems.
• Protein Subcellular Localization: Predicts the subcellular localization of proteins.
• CRISPR/Cas9 sgRNA Activity Prediction: Predicts activity of single-guide RNAs (sgRNAs) in CRISPR/Cas9 systems.

Methodology:

Processes explicitly include text reading, parameter initialization, sequence encoding, model loading, training, and evaluation using deep learning architectures implemented with TensorFlow, PyTorch, and Keras.