SeqInfer is a Python package to infer from sequence, enabling outcome prediction, sequence generation, and meaningful representation discovery, etc for sequence-like data.
Initially focused on biological sequences such as DNA, RNA, and protein sequences, it aims to provide essential tools and algorithms for handling sequence data. However, the package is designed to be easily expandable to accommodate other types of sequences, such as SMILE strings or time series. Relevant helper modules may be added in the future development.
**This library was renamed to SeqInfer from SeqLearn to avoid potential conflicts and confusion given that SeqLearn has been used by other people's repo.
You can install SeqInfer using pip:
pip install seqinfer
However, it is currently recommended to install it directly from git repo given the the pip release may not be up to date in the early stage of development of SeqInfer.
pip install git+https://github.com/jiajiexiao/seqinfer.git
To use SeqInfer, simply import the desired modules from the seq and infer sub-packages.
For example, you can prepare the data as below:
from seqinfer.seq.datasets import SeqFromFileDataset
from seqinfer.seq.transforms import Compose, KmerTokenizer, OneHotEncoder, ToTensor
from seqinfer.seq.vocabularies import unambiguous_dna_vocabulary_dict
seq_dataset = SeqFromFileDataset(
seq_file="examples/toys/CCA-TXXAGG-AG-TGG-TC-A-T/pos.fasta",
seq_file_fmt="fasta",
transform_sequences=Compose(
[
KmerTokenizer(
k=1,
stride=1,
vocab_dict=unambiguous_dna_vocabulary_dict,
num_output_tokens=3,
special_tokens=None,
),
OneHotEncoder(vocab_size=len(unambiguous_dna_vocabulary_dict)),
ToTensor(),
]
),
)The SeqInfer package is organized into two major parts:
seq: Contains modules to define and manage the data/dataset of sequences and provides various related transformation operations.infer: Contains modules for different learners (learning algorithms) to conduct learning tasks such as classification, regression, self-supervised representation learning, sequence generation, etc.
The examples folder contains illustrative examples demonstrating the usage of SeqInfer for various
tasks, including classification, regression, multitask learning, etc. Each example includes a README
to guide you through the usage and expected results.
We welcome contributions to improve and extend SeqInfer. If you would like to contribute, please follow our contribution guidelines (To be added).
This project is licensed under the MIT License - see the LICENSE file for details.
We hope you find SeqInfer useful for your sequence learning tasks! If you encounter any issues or have suggestions for improvement, please feel free to open an issue or submit a pull request. Happy coding!