## Data Sources and Methods
                
### Training Data Collection

Our models are trained on curated datasets from multiple sources:

#### Hemolysis Dataset
- **Primary Source:** [the Database of Antimicrobial Activity and Structure of Peptides (DBAASPv3)](https://academic.oup.com/nar/article-abstract/49/D1/D288/5957160)
- **Secondary Source:** [peptideBERT](https://pubs.acs.org/doi/abs/10.1021/acs.jpclett.3c02398)
- **Size:** 9,316 peptides, with 19.6% being positive (hemolytic) and 80.4% being negative (nonhemolytic)
- **Description:** Probability of peptide disrupting red blood cell membranes.
- **Download:** [hemo-positive.npz](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/hemo-positive.npz)
- **Interpretation** 50% of read blood cells being lysed at x ug/ml concetration (HC50). If HC50 < 100uM, considered as hemolytic, otherwise non-hemolytic.

#### Solubility Dataset
- **Primary Source:** [PROSO-II](https://febs.onlinelibrary.wiley.com/doi/abs/10.1111/j.1742-4658.2012.08603.x)
- **Secondary Source:** [peptideBERT](https://pubs.acs.org/doi/abs/10.1021/acs.jpclett.3c02398)
- **Size:** 18,453 sequences, with 47.6% being labeled as positives and 52.4% being labeled as negatives
- **Description:** Probability of peptide remaining dissolved in aqueous conditions.
- **Download:** [sol-positive.npz](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/sol-positive.npz)

#### Non-Fouling Dataset
- **Primary Source:** [Classifying antimicrobial and multifunctional peptides with Bayesian network models](https://doi.org/10.1002/pep2.24079)
- **Secondary Source:** [peptideBERT](https://pubs.acs.org/doi/abs/10.1021/acs.jpclett.3c02398)
- **Size:** 3,600 positive, 13,585 negative
- **Description:** A nonfouling peptide resists nonspecific interactions and protein adsorption.
- **Download:** [nf-positive.npz](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/nf-positive.npz)

#### Permeability Dataset
- **Primary Source:** [CycPeptMPDB](https://pubs.acs.org/doi/abs/10.1021/acs.jcim.2c01573), [PAMPA](https://doi.org/10.1517/17425255.1.2.325)
- **Secondary Source:** [PepLand](https://arxiv.org/abs/2311.04419)
- **Size:** 1162 positive and negative for nanonical samples each (22 relevant cell-penetrating peptide databases by compiling literature on existing cell-penetrating peptide prediction models ); CycPeptMPDB provides extra 7334 cyclic peptides
- **Description:** Probability of peptide penetrating the cell membrane.
- **Download:** [nc-CPP-processed.csv](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/nc-CPP-processed.csv)

#### Half-life Dataset
- **Primary Source:** [Thpdb2](https://doi.org/10.1016/j.drudis.2024.104047), [PepTherDia](https://doi.org/10.1016/j.drudis.2021.02.019), [peplife](https://www.nature.com/articles/srep36617)
- **Size:** 105 wt, 275 wt+noncanonical, human-only
- **Clean-ups:** Data are all transformed into log\(hour\)
- **Download:** [binding_affinity_training_data.csv](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/half_life_smiles.csv)


#### Binding Affinity Dataset
- **Primary Source:** [PepLand](https://arxiv.org/abs/2311.04419)
- **Size:** 1,781 protein-peptide complexes, canonical and non-canonical
- **Description:** Binding probability normalized in PepLand already. It's a combination of IC50/EC50.
- **Quality:** Binding class cutoffs: Tight ≥ 7.5, Medium 6.0–7.5, Weak < 6.0
- **Download:** [binding_affinity_training_data.csv](https://huggingface.co/ChatterjeeLab/Classifier_Weight/blob/main/training_data/c-binding.csv)
- **Interpretation** Affinity_measure = -log_10(K).

### Model Architecture

- **Sequence Embeddings:** ESM-2 650M parameter model
- **XGBoost Models:** Gradient boosting on pooled ESM embeddings
- **CNN Models:** 1D convolutional networks with attention mechanisms
- **Binding Model:** Cross-attention between protein and peptide representations

### Model Training and Weight Hosting
- [Classifier_weights](https://huggingface.co/ChatterjeeLab/Classifier_Weight)

### 🧪 Physicochemical Properties
                
#### Net Charge Calculation
- Uses Henderson-Hasselbalch equation
- pH-dependent calculation
- Considers all ionizable groups (K, R, H, D, E, C, Y, termini)

#### Isoelectric Point (pI)
- Bisection method to find pH where net charge = 0
- Precision: ±0.01 pH units

#### Hydrophobicity (GRAVY)
- Grand Average of Hydropathy
- Uses Kyte-Doolittle scale
- Range: -4.5 (hydrophilic) to +4.5 (hydrophobic)

### Citation

If you use this tool, please cite:
```
place holder
```

### Contact

For questions or collaborations: [yzhang@u.duke.nus.edu](mailto:yzhang@u.duke.nus.edu)