Sequence-level clonal family assignment, cross-database semantic search, and quantitative repertoire profiling — methodology you can publish on, powered by infrastructure your lab doesn't need to build.
Technical Pipeline
Module-level transparency. Every step is documented, reproducible, and ready for your methods section.
Bulk BCR-seq or 10x single-cell VDJ. FASTQ → adapter trimming, quality filtering, UMI deduplication.
fastp, prestoGermline gene assignment, CDR3 extraction, framework/CDR boundary identification.
IgBLAST, IMGT referenceAntibody-language-model embeddings (AbLang/DNABERT-derived) map each BCR sequence into a 512-d latent space, capturing structural and functional similarity beyond edit distance.
Custom fine-tuned AbLangEmbedding-space clustering assigns clonal families with higher accuracy than Hamming-distance methods. Benchmarked against Change-O/SCOPe on synthetic and real repertoires.
HDBSCAN on embeddingsClonality metrics (Gini, Shannon entropy, top-clone frequency), isotype distribution, SHM quantification, V-gene usage profiling per sample.
Custom feature pipelineCross-database embedding similarity search against IEDB, CoV-AbDab, OAS. Identifies functional homologs even at low sequence identity.
FAISS index, cosine similarityAutoimmune B-Cell Activity Score generation. Per-patient clonal landscape visualization, autoreactive signature flags, stratification tier.
Statistical modelingMethodology
Transformer embeddings separate isotype-switched, affinity-matured clones from naïve populations without manual gating.
Embedding-based clonal family assignment achieves higher V-measure and adjusted Rand index than Hamming-distance clustering (Change-O/SCOPe) on both synthetic lineage trees and paired heavy-light chain validation sets. Full benchmarking data available upon request.
Clono embeddings vs. Hamming-distance (Change-O)
Collaboration
Full data ownership remains with your institution. Clono provides the bioinformatic infrastructure, analysis pipeline, and methodology support — you retain publication rights and intellectual property over your cohort data.
Send us 10–20 samples. We run the full pipeline at no cost and deliver a preliminary report to evaluate fit for your research question.
Joint study design, sample processing, bioinformatic analysis, and co-authorship. We contribute methods expertise; you bring the cohort and clinical context.
Full-service BCR repertoire analysis for your samples with standardized reporting. Ideal for multi-site studies requiring consistent methodology across cohorts.
Selected References
BCR repertoire signatures are disease-specific across 6 immune-mediated diseases (n=209).
Bashford-Rogers R, et al.
BCR repertoire dynamics during B-cell reconstitution correlate with clinical outcomes in RA.
Pollastro S, et al.
scRNA-seq + BCR-seq confirms naïve reconstitution signature equals immune reset in SLE CAR-T.
Wilhelm A, et al.
Pre-treatment autoreactive BCR clones persist through rituximab and dominate at relapse in MuSK-MG.
Stathopoulos P & O'Connor KC
Europe's largest applied research organization. HHI specializes in deep learning and transformer architectures. 76 institutes, €3.4B annual budget.
Co-founded by postdocs in immunology and digital health from Stanford. Deep expertise in B-cell biology, repertoire analysis, and translational biomarker development.
Collaborate
We'll run a complimentary analysis on a small batch of your samples so you can evaluate the methodology firsthand.