bio-dataset-manager 0.1.4

Bio Dataset Manager: easily encode biological sequences into tensors

Authors:

What is it?

Bio Data Manager is a Python project designed for managing and processing bio-sequence data, including DNA, proteins, and SMILES strings. This tool facilitates the encoding of these sequences into tensors, which can then be used for AI computations and complex model implementations.

Project Structure

• bio_data_manager/: Contains core modules for bioinformatics sequence processing and management.
• bio_sequences/: Handles various operations related to biological sequences such as DNA and protein.

Installation

1. Install it as a library
   • Using CPU:

     pip install bio-dataset-manager
     

   • Using CUDA:

     pip install bio-dataset-manager[cuda] -f https://download.pytorch.org/whl/torch_stable.html
     

Usage

Examples of the code can be found in the examples folder.

1. import the modules

import torch
from bio_dataset_manager.bio_dataloader import BioDataloader
from bio_dataset_manager.bio_dataset import BioDataset
from bio_sequences.dna_sequence import DnaSequence

2. create the dataset and dataloader

dataset = BioDataset(
        dataset_folder="path/to/dataset",
        sequences_limit=10,
        randomize_choice=True,
        pad_same_len=False,
        window_size=1,
        sequence_info=DnaSequence(),
        sequences=None,
    )

dataloader = BioDataloader(
        dataset=dataset,
        batch_size=5,
        shuffle=True,
        collate_fn=dataset.collate_fn,
        split_ratio=0.5,
        use_gpu=True if torch.cuda.is_available() else False
    )

3. training loop example

epochs = 5
for epoch in range(epochs):
    with tqdm(total=len(dataloader.training_dataloader), desc=f"Epoch {epoch + 1}/{epochs}", unit="batch") as pbar:
        for batch in dataloader.training_dataloader:
            y_real, lengths = dataloader.process_batch(batch)
            time.sleep(0.1)
            pbar.update(1)
            pbar.set_postfix(
                loss_gen=f"0.0",
                loss_dis=f"0.0"
            )
    pbar.refresh()

Contributing

Feel free to submit issues or pull requests if you'd like to contribute to this project.

License