Sequence Clustering Library
Introduction
This project is currently in development. Feel free to contact me with suggestions for new features or issues.
This project is a Python library designed for clustering biological sequences and splitting datasets into train/test splits while maintaining cluster integrity. It leverages MMseqs2 for sequence clustering and uses pandas for data manipulation.
The library makes it easy to preprocess sequence datasets, cluster them into representative sequences, and create splits that respect clustering boundaries. It’s ideal for bioinformatics workflows where sequence relationships matter.
Vision
The goal of this library is to simplify the preprocessing of biological sequence datasets. I got frustrated by having to switch between python and the command line when performing repetitive preprocessing tasks. While I have started by integrating mmseqs2 with python, the long-term goal of this library is to be a hub of all common biological preprocessing tasks, serving as another level of abstraction on top of existing tools.
The feature list will continue being updated as they are added. Please feel free to reach out with suggested features or post an issue if you find a bug!
Features
-
Sequence Clustering: Use MMseqs2 to cluster biological sequences and add a
representative_sequencecolumn to a DataFrame. -
Train/Test Splitting: Split datasets into train and test sets while ensuring no overlap of sequences across clusters. This maintains the integrity of sequence relationships.