Single-Cell BCR Sequencing

Unlike bulk sequencing, which loses the connection between VH and VL domains, single-cell BCR sequencing physically preserves the cognate pairing of heavy and light chains from each individual B cell. This is critical for downstream antibody structure prediction and antibody-antigen docking, as the native pairing defines the functional paratope. The technique typically uses droplet-based microfluidics or microwell platforms to isolate single cells, followed by overlap extension RT-PCR or template-switching to link the two chains into a single amplicon for sequencing.

Single-cell resolution enables the reconstruction of clonal families by grouping B cells that share a common V(D)J rearrangement and analyzing their somatic hypermutation (SHM) patterns. This phylogenetic analysis reveals the evolutionary trajectory of an antibody response, identifying the unmutated common ancestor (UCA) and intermediate variants. For AI models, these lineage trees provide rich training data for antibody affinity maturation prediction and generative antibody design, as they capture the natural mutation paths that lead to high-affinity binders.

The method captures the full constant region sequence, allowing precise determination of antibody isotype (IgM, IgG, IgA, IgE) and IgG subclass (IgG1, IgG2, IgG3, IgG4). This is essential for developability assessment and Fc engineering, as different isotypes dictate distinct effector functions:

• IgG1: High ADCC and CDC activity
• IgG4: Reduced effector function, preferred for blocking antibodies
• IgA: Mucosal immunity and dimeric secretion Isotype distribution data also informs on the maturation state of the immune response.

Modern single-cell platforms simultaneously capture the BCR transcriptome, the whole transcriptome (gene expression), and surface protein markers (via CITE-seq or TotalSeq) from the same cell. This multi-modal data enables correlation of antibody sequence with the B cell's functional state:

• Plasmablasts: High antibody secretion gene signatures
• Memory B cells: Expression of CD27 and specific homing receptors
• Germinal center B cells: BCL6 expression and active SHM This integration is a cornerstone of multi-omics data integration for AI models predicting antibody function from cellular context.

Commercial platforms can process 10,000 to 100,000 single B cells per sample, generating millions of paired VH:VL sequences in a single experiment. This throughput enables comprehensive immune repertoire sequencing that captures the full diversity of an antigen-specific response, including rare clonotypes. For generative antibody design, these large-scale datasets serve as the foundation for training antibody language models that learn the statistical patterns of natural antibody repertoires, enabling the generation of novel, developable sequences.

Single-cell BCR sequencing is often coupled with fluorescence-activated cell sorting (FACS) using fluorescently labeled antigens to enrich for B cells that bind a specific target. This antigen-baiting approach directly links sequence to function, isolating only the antigen-specific repertoire rather than the bulk background. The resulting datasets are high-quality inputs for antibody-antigen docking and epitope mapping, as every sequenced antibody has a verified binding phenotype. Tetramer-based sorting further enables discrimination of affinity tiers within a single experiment.

The broader category of high-throughput sequencing that profiles the diversity of B-cell receptors (BCRs) or T-cell receptors (TCRs) in a sample. While bulk repertoire sequencing captures aggregate diversity, single-cell BCR sequencing resolves the native pairing of heavy and light chains, preserving the functional unit of the antibody. This paired data is essential for computational antibody discovery because it directly links genotype to binding phenotype without requiring synthetic library assembly.

The computational tracing of point mutations accumulated in antibody variable genes during affinity maturation. Single-cell BCR sequencing captures the clonal lineage of individual B cells, allowing algorithms to reconstruct phylogenetic trees and identify key affinity-conferring residues. This analysis reveals the evolutionary trajectory of an antibody response, pinpointing which mutations drove improved binding and which represent neutral passenger events.

A transformer-based neural network pre-trained on vast repositories of antibody sequences to learn the underlying grammar of immune receptors. These models, such as AntiBERTa or IgLM, are trained on the paired-chain data generated by single-cell BCR sequencing. They learn the co-evolutionary constraints between heavy and light chains, enabling tasks like variant effect prediction, sequence infilling, and the generation of novel antibodies that respect natural pairing rules.

The computational identification of the specific amino acid residues on an antigen that are recognized by an antibody's paratope. Single-cell BCR sequencing enables high-throughput epitope mapping when combined with antigen-specific sorting. By sequencing the paired chains of B cells that bind a known antigen, researchers can cluster antibodies by epitope specificity and infer the structural basis of binding across an entire polyclonal response.

A physics-based or deep learning simulation that predicts the three-dimensional binding pose of an antibody relative to its target antigen. The paired-chain sequences from single-cell BCR sequencing provide the complete variable domain input required for accurate docking. Without native pairing information, docking algorithms must guess the correct heavy-light chain combination, introducing significant uncertainty into the predicted binding interface.

The application of generative models, such as diffusion models and variational autoencoders, to create entirely novel antibody sequences and structures. These models are often conditioned on the paired-chain distributions learned from single-cell BCR sequencing data. By training on native repertoires, generative models produce antibodies that respect the natural heavy-light chain pairing constraints observed in functional immune responses, increasing the likelihood of expressibility and stability.