Sync from GitHub (preserve manual model files)

README.md

@@ -8,15 +8,26 @@ sdk_version: "1.41.1"
 python_version: "3.13"
 app_file: StreamlitApp/StreamlitApp.py
 pinned: false
+short_description: AMP peptide scoring, composition, wheel & 3D views.
 ---
 
 # PeptideAI
-Antimicrobial Peptide (AMP) Prediction App
-A machine learning web app that predicts antimicrobial activity from peptide sequences.
-Built with Python, PyTorch, and Streamlit, it uses ProtBERT embeddings to represent biological sequences and a custom neural network classifier for prediction.
-Includes features for:
 
-- AMP probability prediction
-- Amino acid composition analysis
-- Physicochemical property computation
-- t-SNE visualization of embeddings
+**Live app:** [huggingface.co/spaces/m0ksh/PeptideAI](https://huggingface.co/spaces/m0ksh/PeptideAI)
+
+PeptideAI is a Streamlit app for working with short peptide sequences. It estimates whether a sequence might behave like an antimicrobial peptide (AMP) using a small neural network, and adds views for composition, rough physicochemical numbers, optional mutation search, and helix-style visualization.
+
+## What you can do
+
+- Get an AMP vs non-AMP prediction with a confidence-style score  
+- See amino acid composition and simple properties (length, charge, hydrophobic fraction, mass)  
+- Run a greedy “optimize” pass that tries mutations the model likes more  
+- Visualize a helix-like trace and helical wheel (approximation, not a solved structure)  
+- Run t-SNE on embeddings when you have several sequences  
+
+## Run it on your machine
+
+```bash
+pip install -r requirements.txt
+streamlit run StreamlitApp/StreamlitApp.py
+```

StreamlitApp/StreamlitApp.py

@@ -1,4 +1,4 @@
-# Main Streamlit entrypoint wiring Predict, Analyze, Optimize, Visualize, and t-SNE pages.
+# Main Streamlit entrypoint: one file, several “pages” chosen from the sidebar.
 import streamlit as st
 import pandas as pd
 import numpy as np
@@ -39,7 +39,7 @@ from utils.peptide_extras import (
 )
 
 try:
-    import pyperclip  # Optional; may not exist in all environments.
+    import pyperclip 
 except Exception:
     pyperclip = None
 
@@ -68,17 +68,17 @@ st.divider()
 
 # Initialize session keys so navigation keeps user state across pages.
 if "predictions" not in st.session_state:
-    st.session_state.predictions = []               # list of dicts
+    st.session_state.predictions = []             # list of dicts
 if "predict_ran" not in st.session_state:
     st.session_state.predict_ran = False
 if "predict_input_widget" not in st.session_state:
     st.session_state.predict_input_widget = ""
 if "analyze_input" not in st.session_state:
-    st.session_state.analyze_input = ""             # last analyze input
+    st.session_state.analyze_input = ""           # last analyze input
 if "analyze_output" not in st.session_state:
-    st.session_state.analyze_output = None         # (label, conf_display, comp, props, analysis)
+    st.session_state.analyze_output = None        # (label, conf_display, comp, props, analysis)
 if "optimize_input" not in st.session_state:
-    st.session_state.optimize_input = ""           # last optimize input
+    st.session_state.optimize_input = ""          # last optimize input
 if "optimize_output" not in st.session_state:
     st.session_state.optimize_output = None       # (orig_seq, orig_conf, improved_seq, improved_conf, history)
 if "optimize_last_ran_input" not in st.session_state:
@@ -131,7 +131,7 @@ if st.sidebar.button("Clear All Fields"):
         st.stop()
 
 
-# Cache model weights once per server process for fast repeated inference.
+# Load weights once; every page shares this same model instance.
 model = load_model()
 
 # Shared style tweak keeps expander spacing consistent across pages.
@@ -473,7 +473,7 @@ elif page == "Analyze":
                 mime="text/plain",
             )
 
-# Optimize page: greedy mutation search with per-step diagnostics.
+# Optimize page: Mutation search with per-step diagnostics.
 elif page == "Optimize":
     st.header("Peptide Optimizer")
 
@@ -620,6 +620,7 @@ elif page == "Visualize":
                 st.markdown(f"- {line}")
 
 # t-SNE page: embedding projection for multi-sequence exploration.
+# --- t-SNE on first-layer activations ---
 elif page == "t-SNE":
     st.header("t-SNE Visualizer")
     st.write("Upload peptide sequences (FASTA or plain list) to embed sequences and explore clusters with t-SNE.")
@@ -714,12 +715,14 @@ elif page == "t-SNE":
 • Coloring by properties reveals biochemical trends.
 """)
 
-# About page: quick orientation + disclaimer for new users.
+# --- About (overview + disclaimer) ---
 elif page == "About":
     st.header("About the Project")
     st.markdown("""
 PeptideAI is a lightweight Streamlit app for exploring antimicrobial peptide (AMP) sequences.
 
+A hosted copy may be available at [Hugging Face Spaces](https://huggingface.co/spaces/m0ksh/PeptideAI).
+
 It uses a trained neural network to estimate whether a peptide is likely to be antimicrobial, then helps you interpret and improve candidates:
 - **AMP Predictor**: batch predictions from multi-line or FASTA input, length warnings, persisted results, top-candidate highlight, and CSV export.
 - **Peptide Analyzer**: single-sequence numerical and textual analysis — AMP prediction, composition, physicochemical table + radar, similarity to known AMPs, and report export.

StreamlitApp/utils/analyze.py

@@ -1,4 +1,5 @@
 # Sequence composition and physicochemical property helpers.
+# Mass and charge are textbook approximations for the UI, not for publishing numbers.
 from collections import Counter
 
 def aa_composition(sequence):

StreamlitApp/utils/optimize.py

@@ -1,4 +1,5 @@
 # Heuristic mutation search used by the Optimize page.
+# It’s greedy and uses a few residue buckets — fun to play with, not a real design pipeline.
 import random
 from utils.predict import predict_amp
 

StreamlitApp/utils/peptide_extras.py

@@ -1,5 +1,5 @@
-# Optional peptide UI helpers: 3D approximation (py3Dmol), known-AMP similarity, and residue highlighting.
-# This module is UI-oriented and does not alter model loading or prediction logic.
+# Optional peptide UI helpers: 3D approximation (py3Dmol / Plotly), known-AMP similarity, helical wheel, etc.
+# None of this feeds the classifier — it’s for intuition and teaching.
 from __future__ import annotations
 
 import csv

StreamlitApp/utils/predict.py

@@ -1,4 +1,5 @@
 # Model loading, sequence encoding, and AMP inference helpers.
+# Features are flattened one-hots (length × 20), not transformer embeddings — keeps the app small and CPU-friendly.
 import pathlib
 import numpy as np
 import torch

StreamlitApp/utils/rateLimit.py

@@ -1,4 +1,4 @@
-# Simple in-memory sliding-window rate limiter.
+# Simple in-memory sliding-window rate limiter — good enough for a single Streamlit server.
 import time
 from collections import deque
 

StreamlitApp/utils/ui_helpers.py

@@ -1,4 +1,5 @@
 # UI-facing formatting and summary helpers shared across pages.
+# Lots of HTML strings here — keep them boring and escaped where it matters.
 import html as _html
 from typing import Dict, List, Tuple, Optional
 
@@ -195,6 +196,7 @@ def sequence_health_label(conf_prob: float, charge: float, hydro_fraction: float
     return "Unlikely AMP", "#d62728"
 
 
+# Plain-language bullets for Analyze — rules of thumb, not a second model.
 def build_analysis_insights(
     label: str,
     conf: float,

StreamlitApp/utils/visualize.py

@@ -1,4 +1,4 @@
-# Legacy t-SNE helper retained for ad-hoc embedding previews.
+# t-SNE helper — uses the first linear layer as a quick embedding; main app duplicates this with Plotly inline.
 import pandas as pd
 import matplotlib.pyplot as plt
 from sklearn.manifold import TSNE

requirements.txt

@@ -1,9 +1,9 @@
-streamlit
-pandas
-numpy
-torch
-scikit-learn
-matplotlib
-plotly
-requests
-py3dmol
+streamlit  #1
+pandas  #2
+numpy  #3
+torch  #4
+scikit-learn  #5
+matplotlib  #6
+plotly  #7
+requests  #8
+py3dmol  #9

space_card.yaml

@@ -0,0 +1,12 @@
+# Mirror of the YAML block at the top of README.md (Hugging Face Spaces reads README only).
+
+title: PeptideAI
+emoji: 🔬
+colorFrom: blue
+colorTo: purple
+sdk: streamlit
+sdk_version: "1.41.1"
+python_version: "3.13"
+app_file: StreamlitApp/StreamlitApp.py
+pinned: false
+short_description: AMP peptide scoring, composition, wheel & 3D views.