bionemo-noodles

bionemo-noodles is a Python wrapper of noodles that supports memmap-based file I/O for FASTA files.

Installation

To install from PyPI, execute the following command:

pip install bionemo-noodles

Compatibility

bionemo-noodles has pre-built wheels for Python/Cython 3.10, 3.11, and 3.12, and is compatible with manylinux_2_28 on x86_64.

For a custom build configuration that is not currently supported on PyPI, reach out to: bionemofeedback@nvidia.com

Usage

An example torch.utils.data.Dataset using NvFaidx / bionemo-noodles:

import json
from pathlib import Path

import torch

from bionemo.noodles.nvfaidx import NvFaidx

class SimpleFastaDataset(torch.utils.data.Dataset):

    def __init__(self, fasta_path: Path, tokenizer):
        """Initialize the dataset."""
        super().__init__()
        self.fasta = NvFaidx(fasta_path)
        self.seqids = sorted(self.fasta.keys())
        self.tokenizer = tokenizer

    def write_idx_map(self, output_dir: Path):
        """Write the index map to the output directory."""
        with open(output_dir / "seq_idx_map.json", "w") as f:
            json.dump({seqid: idx for idx, seqid in enumerate(self.seqids)}, f)

    def __len__(self):
        """Get the length of the dataset."""
        return len(self.seqids)

    def __getitem__(self, idx: int) -> dict[str, torch.Tensor]:
        """Get an item from the dataset."""
        sequence = self.fasta[self.seqids[idx]].sequence().upper()
        tokenized_seq = self.tokenizer.text_to_ids(sequence)
        loss_mask = torch.ones_like(torch.tensor(tokenized_seq, dtype=torch.long), dtype=torch.long)
        return {
            "tokens": torch.tensor(tokenized_seq, dtype=torch.long),
            "position_ids": torch.arange(len(tokenized_seq), dtype=torch.long),
            "seq_idx": torch.tensor(idx, dtype=torch.long),
            "loss_mask": loss_mask,
        }

BioNeMo Framework Ecosystem Development

To install this sub-package locally (with --editable):

pip install -e .

To run unit tests, execute:

pytest -v .

To build wheels for different Python, Linux, and system architecture configurations, run the BioNeMo Sub-Package GitHub Actions Workflow (bionemo-subpackage-ci.yml)