clusTCR

clusTCR performs rapid, scalable clustering of CDR3 sequences from T-cell receptors (TCRs) to group similar sequences for analysis of TCR antigen and epitope specificity.

Key Features:

• Two-step clustering: Combines the Faiss Clustering Library for fast initial grouping with the Markov Clustering Algorithm for refined cluster accuracy.
• Scalability: Designed to handle data sets exceeding one million CDR3 sequences and to cluster millions of sequences within minutes.
• Similarity searching and hashing: Uses similarity searching and sequence hashing to accelerate identification of related TCR sequences.
• CDR3-focused analysis: Specifically targets CDR3 regions of TCRs for clustering relevant to antigen recognition.
• Benchmark performance: Demonstrates similar accuracy to existing TCR clustering methods while providing substantial speed improvements.

Scientific Applications:

• Large-scale repertoire analysis: Clustering millions of CDR3 sequences to characterize TCR repertoire diversity and structure.
• Epitope specificity exploration: Grouping TCRs by similarity to support identification of antigen- or epitope-associated clusters.
• Scalable immunological studies: Enabling high-throughput analysis of large immunological sequencing data sets for studies of T-cell specificity.

Methodology:

Employs a two-step computational approach using Faiss for fast similarity searching and sequence hashing followed by the Markov Clustering Algorithm for cluster refinement.