🧬 Protein Sequences Classification

This project explores the application of Machine Learning (ML), Deep Learning (DL), and Transfer Learning (TL) techniques to classify structural protein sequences. By treating protein sequences like natural language, we leverage NLP-inspired approaches to identify and classify structural types based on sequence patterns.

---

📌 Project Overview

Proteins are composed of sequences of amino acids, and their structure plays a critical role in their function. This project applies traditional ML algorithms, sequence-based neural networks, and state-of-the-art pretrained models for classifying protein sequences into structural categories.

---

🧠 Model Architectures

🔹 Machine Learning Models

• Naive Bayes
• XGBoost
• Logistic Regression
• K-Nearest Neighbors (KNN)

🔹 Deep Learning Models

• Bidirectional LSTM (BiLSTM)
• Convolutional Neural Network (CNN)
• Recurrent Neural Network (RNN)
• Gated Recurrent Unit (GRU)

🔹 Transfer Learning Models

• ProtBERT
• ESM2

---

📊 Evaluation Metrics

🔹 Machine Learning Results

Model	Accuracy
Naive Bayes	86.53%
XGBoost	81.26%
Logistic Regression	90.45%
KNN	75.54%

---

🔹 Deep Learning Results

Model	Accuracy
BiLSTM	84.19%
CNN	82.43%
GRU	84.71%
RNN	50.67%

---

🔹 Transfer Learning Results

Model	Accuracy
ProtBERT	82.79%
ESM2	96.43%

---

🧪 Dataset

• Source: Kaggle - Protein Dataset
• Classes: 8
• Size: 67728 sequences

---

🛠️ Tech Stack

• Python
• NumPy
• Pandas
• Scikit-learn
• XGBoost
• PyTorch
• Transformers (Hugging Face)

---

📫 Stay in touch

• Author - Naman Arora
• Twitter - @naman_22a

🗒️ License

Protein Sequences Classification is licensed under GPL V3