Immune Repertoire Sequencing

A high-resolution technique that captures paired heavy- and light-chain sequences from individual B cells, preserving the native cognate pairing information lost in bulk sequencing.

• Enables direct interrogation of functional antibody units
• Critical for identifying rare, potent neutralizing antibodies
• Generates training data for antibody language models
• Platforms: 10x Genomics, BD Rhapsody, Berkeley Lights

Unlike bulk repertoire sequencing, single-cell methods resolve the exact VH-VL pairing, which is essential for recombinant antibody expression and functional validation.

The computational tracing of point mutations accumulated in antibody variable genes during the affinity maturation process within germinal centers.

• Reconstructs B-cell lineage trees from repertoire data
• Identifies key affinity-conferring residues under positive selection
• Distinguishes antigen-driven selection from background mutation
• Tools: IgPhyML, BASELINe, SHAZAM

This analysis reveals the evolutionary pathways that produced high-affinity antibodies, informing machine learning models that predict the effect of novel mutations on binding strength.

A transformer-based neural network pre-trained on vast repositories of antibody sequences to learn the underlying grammar and constraints of immune receptors.

• Learns residue co-evolution patterns across millions of natural repertoires
• Used for variant effect prediction and sequence likelihood scoring
• Can generate novel, physically plausible antibody sequences
• Examples: AntiBERTy, AbLang, ESM-IF for antibodies

These models capture the complex dependencies between positions that govern folding and function, enabling zero-shot prediction of mutations that enhance developability or affinity.

The application of generative models—including diffusion models, variational autoencoders, and autoregressive transformers—to create entirely novel antibody sequences and structures beyond natural repertoire screening.

• Samples from learned distributions of functional antibody space
• Conditions generation on target epitope or desired properties
• Explores sequence diversity not present in natural repertoires
• Frameworks: DiffAb, IgVAE, RefineGNN

This approach moves beyond mining existing repertoires to de novo creation of binders with optimized affinity, specificity, and developability profiles.

A high-throughput experimental technique that measures the functional effect of thousands of single amino acid substitutions in an antibody, generating rich datasets for training supervised variant effect predictors.

• Couples mutagenesis libraries with functional selection assays
• Quantifies the fitness landscape of every position in the sequence
• Provides ground-truth labels for supervised machine learning
• Enables training of models that predict mutation effects without wet-lab screening

These datasets bridge repertoire sequencing observations with functional validation, enabling accurate in silico prediction of affinity-enhancing mutations.

A computational framework that simultaneously optimizes an antibody sequence for multiple, often conflicting properties—such as affinity, specificity, solubility, and immunogenicity—to identify Pareto-optimal designs.

• Balances trade-offs between binding potency and developability
• Uses Bayesian optimization or evolutionary algorithms
• Integrates predictions from multiple surrogate models
• Outputs a frontier of non-dominated candidate sequences

Repertoire sequencing data provides the natural sequence constraints that ground these optimization algorithms, ensuring generated designs remain within physically realizable antibody space.