Update README.md

README.md

@@ -17,7 +17,7 @@ This model is a fine-tuned version of ESM2-3B for paired antibody sequences (hea
 
 **Key Features:**
 - Trained on paired antibody sequences
-- 15% WC followed by 50% CDR fine-tuning
+- 15% whole-chain (WC) masking followed by 50% CDR masking during fine-tuning
 - Input format: Heavy-Light chains separated by "-"
 - Output: 2560-dimensional embeddings
 - Optimized for antibody CDR region understanding
@@ -59,7 +59,7 @@ light_chain = (
 paired_sequence = f"{heavy_chain}{SEP_TOKEN}{light_chain}"
 
 # Tokenize
-inputs = tokenizer(paired, return_tensors="pt", add_special_tokens=True)
+inputs = tokenizer(paired_sequence, return_tensors="pt", add_special_tokens=True)
 
 # Extract embeddings
 with torch.no_grad():
@@ -93,17 +93,34 @@ sequence = "EVQLVESGGGLVQPGGSLRLSCAASGFTFSSYAMS...-DIQMTQSPSSLSASVGDRVTITCRASQSI
 - **Sequence length:** Variable (up to ~1024 tokens including special tokens)
 - **Format:** PyTorch tensor
 
+## Publication
+
+**Preferential CDR masking in paired antibody language models improves binding affinity prediction**
+
+Mahtab Talaei, Kenji C. Walker, Boran Hao, Eliot Jolley, Yeping Jin, Dima Kozakov, John Misasi, Sandor Vajda, Ioannis Ch. Paschalidis, Diane Joseph-McCarthy.
+
+*Communications AI & Computing* **1**, 7 (2026).
+
+**DOI:** https://doi.org/10.1038/s44488-026-00010-2
+
+**Published:** August 13, 2026
+
+[Read the published article](https://www.nature.com/articles/s44488-026-00010-2)
+
+
 ## Citation
 
-If you use this model, please cite:
+If you use this model, please cite the published paper:
 
 ```bibtex
-@article{talaei2025preferential,
-  title={Preferential CDR masking in paired antibody language models improves binding affinity prediction},
-  author={Talaei, Mahtab and Walker, Kenji C. and Hao, Boran and Jolley, Eliot and Jin, Yeping and Kozakov, Dima and Misasi, John and Vajda, Sandor and Paschalidis, Ioannis Ch. and Joseph-McCarthy, Diane},
-  journal={bioRxiv},
-  year={2025},
-  doi={10.1101/2025.10.31.685149}
+@article{Talaei2026,
+  author = {Talaei, Mahtab and Walker, Kenji C. and Hao, Boran and Jolley, Eliot and Jin, Yeping and Kozakov, Dima and Misasi, John and Vajda, Sandor and Paschalidis, Ioannis Ch. and Joseph-McCarthy, Diane},
+  title = {Preferential {CDR} masking in paired antibody language models improves binding affinity prediction},
+  journal = {Communications AI \& Computing},
+  volume = {1},
+  pages = {7},
+  year = {2026},
+  doi = {10.1038/s44488-026-00010-2}
 }
 ```
 
@@ -112,7 +129,7 @@ If you use this model, please cite:
 - **Maintainer:** Network Optimization & Control (NOC) Lab
 - **Email:** mtalaei@bu.edu
 - **GitHub:** [https://github.com/noc-lab/AbCDR-ESM](https://github.com/noc-lab/AbCDR-ESM)
-- **Paper:** [bioRxiv preprint](https://www.biorxiv.org/content/10.1101/2025.10.31.685149)
+- **Paper:** [Communications AI & Computing](https://doi.org/10.1038/s44488-026-00010-2)
 
 ## License