GATHeR: graph-based accurate tool for immunoglobulin heavy- and light-chain reconstruction

Abstract

Abstract Recovering full-length, paired B-cell receptor (BCR) sequences from scRNA-seq reads remains difficult, especially in naive and memory B cells where immunoglobulin transcripts are sparse. Incomplete constant-region coverage in current methods limits isoform, subclass, and allele resolution. Here we present GATHeR, an open-source tool that assembles and annotates paired heavy- and light-chain BCR sequences and extends assembled sequences into constant regions. This enables confident subclass and allele assignment and recovery of membrane-bound isoforms, including the transmembrane segment and cytoplasmic tail, thereby distinguishing surface BCRs from secreted antibodies. GATHeR supports Smart-seq2/3 and 10x Genomics libraries and outperforms existing methods across benchmarks, with the largest gains in naive and memory B cells. Notably, in these populations the constant-region extension also enables detection of splice variation, including intron-containing heavy-chain transcripts with read-level support. By delivering high-fidelity receptor, isoform, and clonal lineage information, GATHeR broadens the analytical reach of scRNA-seq for B-cell immunology.