Update app.py
Browse files
app.py
CHANGED
|
@@ -1,77 +1,104 @@
|
|
| 1 |
import streamlit as st
|
| 2 |
-
import
|
| 3 |
-
import hashlib
|
| 4 |
import py3Dmol
|
| 5 |
import requests
|
| 6 |
-
import io
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
|
|
|
| 9 |
def generate_sequence_from_words(words, length):
|
| 10 |
seed = ' '.join(words).encode('utf-8')
|
| 11 |
random.seed(hashlib.md5(seed).hexdigest())
|
| 12 |
amino_acids = "ACDEFGHIKLMNPQRSTVWY"
|
| 13 |
return ''.join(random.choice(amino_acids) for _ in range(length))
|
| 14 |
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
return None
|
| 26 |
-
|
| 27 |
-
def visualize_protein(pdb_string):
|
| 28 |
-
view = py3Dmol.view(width=800, height=400)
|
| 29 |
-
view.addModel(pdb_string, 'pdb')
|
| 30 |
-
view.setStyle({'cartoon': {'color': 'spectrum'}})
|
| 31 |
-
view.zoomTo()
|
| 32 |
-
return view
|
| 33 |
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
word2 = st.text_input("Word 2")
|
| 39 |
-
word3 = st.text_input("Word 3")
|
| 40 |
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
|
|
|
| 46 |
|
| 47 |
-
|
|
|
|
| 48 |
if word1 and word2 and word3:
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
st.
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
st.header("Protein Structure Prediction")
|
| 55 |
with st.spinner("Predicting protein structure... This may take a few minutes."):
|
| 56 |
-
|
| 57 |
-
if pdb_string:
|
| 58 |
-
view = visualize_protein(pdb_string)
|
| 59 |
-
|
| 60 |
-
st_py3dmol = py3Dmol.show3d(view, width=800, height=400)
|
| 61 |
-
st.components.v1.html(st_py3dmol.startjs, height=400)
|
| 62 |
-
|
| 63 |
-
st.success("Structure prediction complete!")
|
| 64 |
-
st.write("Note: This is a computational prediction and may not represent the actual biological structure.")
|
| 65 |
-
else:
|
| 66 |
-
st.error("Failed to predict structure. Please try again.")
|
| 67 |
else:
|
| 68 |
-
st.
|
| 69 |
|
| 70 |
-
|
|
|
|
| 71 |
## What to do next:
|
| 72 |
-
1.
|
| 73 |
-
2.
|
| 74 |
-
3.
|
| 75 |
-
4.
|
| 76 |
-
|
|
|
|
| 77 |
""")
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
+
from stmol import showmol
|
|
|
|
| 3 |
import py3Dmol
|
| 4 |
import requests
|
| 5 |
+
import biotite.structure.io as bsio
|
| 6 |
+
import random
|
| 7 |
+
import hashlib
|
| 8 |
+
import urllib3
|
| 9 |
+
|
| 10 |
+
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
|
| 11 |
+
|
| 12 |
+
st.set_page_config(layout='wide')
|
| 13 |
+
st.sidebar.title('🎈 ESMFold Protein Structure Predictor')
|
| 14 |
+
st.sidebar.write('[*ESMFold*](https://esmatlas.com/about) is an end-to-end single sequence protein structure predictor based on the ESM-2 language model. For more information, read the [research article](https://www.biorxiv.org/content/10.1101/2022.07.20.500902v2) and the [news article](https://www.nature.com/articles/d41586-022-03539-1) published in *Nature*.')
|
| 15 |
|
| 16 |
+
# Function to generate protein sequence from words
|
| 17 |
def generate_sequence_from_words(words, length):
|
| 18 |
seed = ' '.join(words).encode('utf-8')
|
| 19 |
random.seed(hashlib.md5(seed).hexdigest())
|
| 20 |
amino_acids = "ACDEFGHIKLMNPQRSTVWY"
|
| 21 |
return ''.join(random.choice(amino_acids) for _ in range(length))
|
| 22 |
|
| 23 |
+
# stmol
|
| 24 |
+
def render_mol(pdb):
|
| 25 |
+
pdbview = py3Dmol.view()
|
| 26 |
+
pdbview.addModel(pdb,'pdb')
|
| 27 |
+
pdbview.setStyle({'cartoon':{'color':'spectrum'}})
|
| 28 |
+
pdbview.setBackgroundColor('white')
|
| 29 |
+
pdbview.zoomTo()
|
| 30 |
+
pdbview.zoom(2, 800)
|
| 31 |
+
pdbview.spin(True)
|
| 32 |
+
showmol(pdbview, height = 500,width=800)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
+
# ESMfold
|
| 35 |
+
def update(sequence, word1, word2, word3, sequence_length):
|
| 36 |
+
headers = {
|
| 37 |
+
'Content-Type': 'application/x-www-form-urlencoded',
|
| 38 |
+
}
|
| 39 |
+
try:
|
| 40 |
+
response = requests.post('https://api.esmatlas.com/foldSequence/v1/pdb/',
|
| 41 |
+
headers=headers,
|
| 42 |
+
data=sequence,
|
| 43 |
+
verify=False, # Disable SSL verification
|
| 44 |
+
timeout=300) # Set a longer timeout
|
| 45 |
+
response.raise_for_status() # Raise an exception for bad status codes
|
| 46 |
+
pdb_string = response.content.decode('utf-8')
|
| 47 |
+
|
| 48 |
+
with open('predicted.pdb', 'w') as f:
|
| 49 |
+
f.write(pdb_string)
|
| 50 |
+
|
| 51 |
+
struct = bsio.load_structure('predicted.pdb', extra_fields=["b_factor"])
|
| 52 |
+
b_value = round(struct.b_factor.mean(), 2)
|
| 53 |
+
|
| 54 |
+
# Display protein structure
|
| 55 |
+
st.subheader(f'Predicted protein structure using seed: {word1}, {word2}, and {word3} + length ({sequence_length})')
|
| 56 |
+
render_mol(pdb_string)
|
| 57 |
+
|
| 58 |
+
# plDDT value is stored in the B-factor field
|
| 59 |
+
st.subheader('plDDT Score')
|
| 60 |
+
st.write('plDDT is a per-residue estimate of the confidence in prediction on a scale from 0-100.')
|
| 61 |
+
st.info(f'Average plDDT: {b_value}%')
|
| 62 |
+
|
| 63 |
+
st.download_button(
|
| 64 |
+
label="Download PDB",
|
| 65 |
+
data=pdb_string,
|
| 66 |
+
file_name='predicted.pdb',
|
| 67 |
+
mime='text/plain',
|
| 68 |
+
)
|
| 69 |
+
except requests.exceptions.RequestException as e:
|
| 70 |
+
st.error(f"An error occurred while calling the API: {str(e)}")
|
| 71 |
+
st.write("Please try again later or contact support if the issue persists.")
|
| 72 |
|
| 73 |
+
# Streamlit app
|
| 74 |
+
st.title("Word-Seeded Protein Sequence Generator and Structure Predictor")
|
|
|
|
|
|
|
| 75 |
|
| 76 |
+
# Input for word-seeded sequence generation
|
| 77 |
+
st.sidebar.subheader("Generate Sequence from Words")
|
| 78 |
+
word1 = st.sidebar.text_input("Word 1")
|
| 79 |
+
word2 = st.sidebar.text_input("Word 2")
|
| 80 |
+
word3 = st.sidebar.text_input("Word 3")
|
| 81 |
+
sequence_length = st.sidebar.number_input("Sequence Length", min_value=50, max_value=400, value=100, step=10)
|
| 82 |
|
| 83 |
+
# Generate and predict button
|
| 84 |
+
if st.sidebar.button('Generate and Predict'):
|
| 85 |
if word1 and word2 and word3:
|
| 86 |
+
sequence = generate_sequence_from_words([word1, word2, word3], sequence_length)
|
| 87 |
+
st.sidebar.text_area("Generated Sequence", sequence, height=100)
|
| 88 |
+
st.sidebar.info("Note: The same words and length will always produce the same sequence.")
|
| 89 |
+
|
|
|
|
|
|
|
| 90 |
with st.spinner("Predicting protein structure... This may take a few minutes."):
|
| 91 |
+
update(sequence, word1, word2, word3, sequence_length)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 92 |
else:
|
| 93 |
+
st.sidebar.warning("Please enter all three words to generate a sequence.")
|
| 94 |
|
| 95 |
+
# Information display
|
| 96 |
+
st.sidebar.markdown("""
|
| 97 |
## What to do next:
|
| 98 |
+
1. Enter three words and a sequence length.
|
| 99 |
+
2. Click 'Generate and Predict' to generate the sequence, visualize the protein, and get its plDDT score.
|
| 100 |
+
3. Explore the 3D structure and download the PDB file if desired.
|
| 101 |
+
4. Experiment with different words or sequence lengths to see how they affect the predicted structure.
|
| 102 |
+
|
| 103 |
+
Remember, these predictions are based on AI models and should be interpreted with caution.
|
| 104 |
""")
|