Computational Antibody Papers
Filter by tags
All
Filter by published year
All
TitleKey points
2026-03-27
Latent-Y: A Lab-Validated Autonomous Agent for De Novo Drug Design
- nanobodies
- protein design
- agents
- First autonomous nanobody design agent.
- Prompted by high-level goals: It translates natural language objectives, like "inhibit X interaction with Y", into complete design campaigns.
- The agent queries literature/databases, uses bioinformatics tools, and prompts the user for specific strategic clarifications.
- 56x expert-level speedup, by compressing weeks of expert research and computational tasks into hours by automating reasoning-intensive steps.
- In lab tests, it successfully generated functional binders for 6 out of 9 attempted targets.
- Authors advocate for a "prompt-to-drug" autonomous pipeline, using a central AI orchestrator to connect disparate pre-clinical and clinical steps agentically.
- While modular proofs-of-concept exist, they remain domain-specific, brittle, and far from full-cycle implementation in actual drug discovery programs.
- A primary recommendation is to eliminate "data silos" by making research open, peer-reviewed, and accessible via APIs to ensure outputs are easily "machine-readable" for AI training.
- The system faces significant hurdles from LLM hallucinations and "cascading errors," where a single early-stage miscalculation (like an incorrect binding pocket) propagates through the entire chain.
- Despite the push for autonomy, authors argue "human-in-the-loop" checkpoints remain legally and ethically mandatory for high-stakes regulatory and clinical transitions.
2026-03-27
Application of protein language models for antibody developability prediction
- developability
- Analysis of developability data from 33 internal Biogen programs, covering 18,540 antibodies.
- Focused on three dimensions: hydrophobicity (HIC), polyspecificity (PSR), and self-association (AC-SINS).
- Labeled subsets included 4,594 (PSR), 1,792 (HIC), and 7,727 (AC-SINS) sequences.
- Benchmarked three PLMs: ESM2 (general-purpose), plus IgBert and IgT5 (antibody-specific).
- Domain-adaptive fine-tuning consistently boosted antibody-specific PLMs, but often degraded ESM2 performance.
- Antibody-specific PLMs generally provided better embeddings for PSR and AC-SINS, while ESM2 remained highly competitive for HIC.
- Perplexity was only weakly correlated in aggregate, but showed significant association with PSR/AC-SINS failure when controlled for a fixed light chain
- Novel experimental and computational pipeline designed to characterize nanobody immune repertoires following immunization and phage display selection - NanoMAP.
- It introduces a flexible clustering method that identifies clonal families by grouping sequences with similar V/J segments and CDR lengths, then applying a unique merging step that allows for minor CDR variations.
- When benchmarked against MMseqs2 and Immcantation (SCOPer), NanoMAP scored higher on computational metrics (Silhouette, phenotypic quality, and stability) and showed better alignment with expert-curated "ground truth" labels.
2026-03-20
Scaling Atomistic Protein Binder Design with Generative Pretraining And Test-time Compute
- protein design
- Novel generative framework to design protein binders from NVIDIA.
- Antibodies/nanobodie are not singled out for analysis.
- First framework to unify generative modeling with hallucination-based optimization, allowing for a strong generative prior to be steered by inference-time compute.
- The authors introduced Teddymer, a dataset of ~510,000 synthetic dimers created from AlphaFold predicted domain-domain interactions to overcome the scarcity of experimental multimer data.
- The model uses advanced search algorithms, including Beam Search, Feynman-Kac Steering, and MCTS, to navigate the generative space and find high-quality binders.
- It achieved state-of-the-art results on protein targets, small molecules, and enzyme design tasks, consistently outperforming baselines like RFDiffusion and BindCraft.
- No Wet-Lab testing. Hopefully just yet.
2026-03-20
Programming Biomolecular Interactions with All-Atom Generative Model
- nanobodies
- protein design
- AnewOmni, foundation model that unifies the design of small molecules, peptides, and antibodies into a single framework.
- The team evaluated approximately 3,000 candidates for the "undruggable" KRAS G12D target by alternating between AnewOmni for CDR design and AlphaFold3 for structural validation.
- Out of 7 synthesized nanobodies, the model achieved a 75% success rate (3 out of 4) when using a conservative structural consistency filter.
- The most successful nanobody design demonstrated a high binding affinity with a Kd of 587 nM
2026-03-13
Contrastive learning for antibody-antigen sequence-to-specificity prediction
- binding prediction
- CALM, a "sequence-native" foundation model that maps antibody and antigen primary sequences without requiring structural inference.
- CALM employs modality-specific encoders (AntiBERTy for antibodies, ESM-2 for antigens) to align cognate pairs in a shared embedding space using cosine similarity.
- Authors evaluate performance by the model's ability to pick the correct partner from a candidate pool in both directions ab->ag, ag->ab.
- Calm uses optional structural masks to restrict inputs to paratope and epitope residues, which significantly reduces sequence noise and improves accuracy (but clearly needs a structure).
- CALM achieves Top-1 of 2% in strict out-of-distribution tests, representing a 3x to 46x improvement over random baselines despite a low-data regime.
- They lay out an autoregressive decoder for de novo design, though this generative component was not trained or tested in this study.
2026-03-13
Structural Plausibility Without Binding Specificity: Limits of AI-Based Antibody-Antigen Structure Prediction Confidence Scores
- binding prediction
- nanobodies
- structure prediction
- The authors evaluated AlphaFold3, Boltz-2, and Chai-1 on their ability to distinguish cognate (correct) nanobody-antigen pairs from incorrect, non-binding pairings.
- They used 106 experimental complexes and generated a combinatorial matrix of 11,132 shuffled non-cognate pairings to serve as ground-truth "incorrect" decoys.
- Internal confidence scores (specifically ipTM) were very weakly predictive of true binding. In terms of Average Precision (PR-AUC), AF3 performed best, followed by Chai-1 and then Boltz-2.
- Increased sampling improves structural geometry but does not help models "select" the correct binder. Most quality gains occur within 10–25 samples; deeper sampling primarily increases the number of plausible-looking false positives.
- Novel training scheme for antibody language models, modeling phylogenetic relationships rather than pure mutational MLM - called DASM.
- Unlike AbLang2’s standard masked language modeling , DASM uses a mutation-selection framework that factors out nucleotide-level biases (like the codon table and SHM rates) to isolate purely functional selection effects.
- The model was trained on approximately 2 million parent-child sequence pairs derived from reconstructed B cell phylogenies , using datasets such as JaffePaired, Tang, and Vanwinkle.
- Model is a compact 4-million-parameter Transformer-encoder featuring 5 layers, 8 attention heads , and a custom "wiggle" activation function to stabilize output selection factors.
- DASM was validated on the FLAb collection (Koenig and Shanehsazzadeh datasets) and MAGMA-seq high-throughput binding assays for influenza and SARS-CoV-2 antibodies. It was better than ABlang2, progen2 and esm2.
2026-02-26
Disulphide and sequence-encoded conformational priors guide nanobody structure prediction
- nanobodies
- structure prediction
- Authors propose a new method to train a nanobody structure predictor by using ‘blueprints’.
- They developed a classifier (NbFrame) to identify whether the HCDR3 loop adopts a kinked (framework-contacting) or extended (solvent-exposed) conformation. This allows the model to use sequence-encoded priors and explicit constraints during the folding process.
- The model itself is very lightweight and runs significantly faster than "heavy" models like AF or Boltz. NbForge achieves sub-second inference speeds, predicting structures in less than a second on both CPU and GPU. In comparison, models like AlphaFold3 or Boltz1 typically require tens of seconds to minutes per structure - MSA is a different story altogether.
- The model matches the HCDR3 prediction quality of heavy models while being much more efficient. While AF3 and Boltz1 are more accurate at modeling the rigid framework, NbForge achieves parity in the hypervariable HCDR3 region—the part most critical for binding. Its speed and high recovery of disulphide bonds make it ideal for triaging millions of candidates in large-scale discovery campaigns.