AbAffinity: A Large Language Model for Predicting Antibody Binding Affinity against SARS-CoV-2
Authors: Faisal Bin Ashraf, Animesh Ray, Stefano Lonardi
Organizations: Department of Computer Science and Engineering,University of California, Riverside, 92521, CA, United States · Riggs School of Applied Life Sciences, Keck Graduate Institute, Claremont, 91711, CA, United States
Machine learning-based antibody design is emerging as one of the most promising approaches to combat infectious diseases, due to significant advancements in the field of artificial intelligence and an exponential surge in experimental antibody data (in particular related to COVID-19). The ability of an antibody to bind to an antigens (called binding affinity) is one of the the most critical properties in designing neutralizing antibodies. In this study we introduce Ab-Affinity, a new large language model that can accurately predict the binding affinity of antibodies against a target peptide, e.g., the SARS-CoV-2 spike protein. Code and model are available at https://github.com/ucrbioinfo/AbAffinity.
Figures & tables
Figure 1: Overview of Ab-Affinity. Ab-Affinity predicts the binding affinity against a specific target peptide; it can also provide residue-residue contact maps and an embedding of the input sequence.
Figure 2: Dataset description. a. Distribution of Binding Affinity ( logKD ); b. Distribution of antibodies from each seed; c. Distribution of antibodies with mutation.
Figure 3: Model Architecture of Ab-Affinity
Figure 4: Comparing affinity prediction models: (a) Pearson and Spearman correlation for DG-Affinity (p-values = 3.86×10−14 , and 3.38×10−15 ), ESM-2 (p-values = 8.03×10−198 , and 8.02×10−198 ), AbLang (p-values = 1.24×10−175 , and 1.02×10−163 ) and Ab-Affinity (p-values = 4.03×10−261 , and 8.65×10−217 ); scatter plot for actual vs. predicted binding affinity for (b) DG-Affinity, (c) ESM-2, (d) AbLang, (e) Ab-Affinity (includes antibodies for all three seeds)
Figure 5: t-SNE representation of the embedding produced by ESM-2 and Ab-Affinity; antibodies are colored according to their predicted binding affinity.
Model
Ref
14H
14L
Pearson
Spearman
Pearson
Spearman
Ens-Grad
[ 10 ]
0.601
0.476
0.637
0.645
ESM-F
[ 6 ]
0.634
0.516
0.674
0.681
AntiBERTa2
[ 2 ]
0.623
0.545
0.673
0.684
AbMAP
[ 17 ]
0.606
0.510
0.674
0.685
A2Binder
[ 6 ]
0.642
0.553
0.683
0.688
Table 1: Comparing Ab-Affinity against other binding affinity predictors on 14H and 14L dataset
Figure 6: ROC curves and AUC values for two classification tasks, namely (a,b) determining the binding affinity class of an antibody (High, Medium, Low) and (c,d) determining whether the binding is improved compared to the seed antibody (Y/N); (a,c) ROC curves using the ESM-2 embedding; (b,d) ROC curves using the Ab-Affinity embedding; AUC values are reported for each classifier
Figure 7: Differences between the attention residue-residue maps for strong binding (i.e. logKd<0.5 ) and weak binding (i.e., logKd>5.5 ) for antibodies generated from Ab-14 (a) heavy chain, and (b) light chain.
Figure 8: t-SNE visualization of the embeddings produced by ESM-2 and Ab-Affinity; points are colored according to the experimentally-determined thermostability of those antibodies
Accurate prediction of antibody-antigen binding affinity is fundamental to therapeutic design, yet remains constrained by severe label sparsity and the complexity of antigenic variations. In this paper, we propose AbLWR (Antibody-antigen binding affinity List-Wise Ranking), a novel framework that reformulates the conventional affinity regression task as a listwise ranking problem. To mitigate label sparsity, AbLWR incorporates a PU (Positive-Unlabeled) learning mechanism leveraging a dual-level contrastive objective and meta-optimized label refinement to learn robust representations. Furthermore, we address antigenic variation by employing a homologous antigen sampling strategy where Multi-Head Self-Attention (MHSA) explicitly models inter-sample relationships within training lists to capture subtle affinity nuances. Extensive experiments demonstrate that AbLWR significantly outperforms state-of-the-art baselines, improving the Precision@1 (P@1) by over 10% in randomized cross-validation experiments. Notably, case studies on Influenza and IL-33 validate its practical utility, demonstrating robust ranking consistency in distinguishing subtle viral mutations and efficiently prioritizing top-tier candidates for wet-lab screening.
Fan Xu, Zhi-an Huang, Haohuai He +5
Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University, Hong Kong SAR, China · Department of Computer Science, City University of Hong Kong (Dongguan), Dongguan, China · School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, Guangdong, China +1
DuaDeep-SeqAffinity is a sequence-only deep learning framework that predicts antibody--antigen binding affinity directly from primary amino acid sequences, avoiding the cost and scarcity of resolved three-dimensional structures. The antigen and the antibody heavy and light chains are processed as three independent streams, each embedded with a frozen ESM-2 protein language model and passed through parallel Transformer and convolutional neural network (CNN) branches before late fusion, a decoupled design intended to preserve local complementarity-determining region (CDR) signal that monolithic encoders can dilute. On a sequence-disjoint split of the AbRank benchmark, the model achieves a Pearson correlation of 0.683, an R^2 of 0.460, and a pairwise ranking AUC of 0.895, significantly outperforming single-branch ablations (paired t-test, p < 0.05). Attention-map and gradient-based saliency analyses further show that the model preferentially attends to CDR loops and candidate epitope residues, supporting its use as a scalable, structure-free tool for high-throughput antibody screening.
Aicha Boutorh, Soumia Bouyahiaoui, Manel Kara Laouar +3
National School of Artificial Intelligence (ENSIA) Sidi Abdallah Campus, Algiers, Algeria · Lamine Debaghine University Hospital Center, CHU Bab El Oued, Algiers, Algeria
Antibodies are essential proteins that play a central role in immune recognition by binding specific antigen molecules. Although recent protein language models have enabled progress in single-chain protein modeling and generation, they often fall short in antigen-specific antibody design, where effective modeling requires explicit pairing between antibody and antigen, particularly at the epitope level. To address these limitations, we introduce AAMFM, an Antigen-specific Antibody Multimodal Foundation Model that learns unified representations of antibody sequences and structures conditioned on antigen context. AAMFM incorporates rich antigen information including geometric interfaces and epitope annotations via a cross-modal adapter, enabling joint modeling of antibody-antigen interactions in a shared latent space. To further guide the model toward functional relevance, we fine-tune AAMFM using Calibrated Direct Preference Optimization (Cal-DPO), leveraging preference signals extracted from a strong structural prior to align learning with binding-specific objectives. Extensive experiments demonstrate that AAMFM achieves state-of-the-art performance in functional antibody design, revealing its potential for antigen-specific antibody engineering. Our code is available at https://github.com/XL-S224/AAMFM.