Atom Bioworks
commited on
Create gui.py
Browse files
gui.py
ADDED
|
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from api_prediction import AptaTransPipeline_Dist
|
| 2 |
+
import gradio as gr
|
| 3 |
+
import pandas as pd
|
| 4 |
+
import torch
|
| 5 |
+
import tempfile
|
| 6 |
+
from tabulate import tabulate
|
| 7 |
+
from PIL import Image
|
| 8 |
+
import itertools
|
| 9 |
+
import os
|
| 10 |
+
import RNA
|
| 11 |
+
import matplotlib.pyplot as plt
|
| 12 |
+
import matplotlib.image as mpimg
|
| 13 |
+
import random
|
| 14 |
+
from scipy.cluster.hierarchy import dendrogram, linkage
|
| 15 |
+
# Visualization
|
| 16 |
+
from Bio.Phylo.PhyloXML import Phylogeny
|
| 17 |
+
from Bio import SeqIO
|
| 18 |
+
from Bio.Seq import Seq
|
| 19 |
+
from Bio.SeqRecord import SeqRecord
|
| 20 |
+
from Bio import AlignIO
|
| 21 |
+
from Bio.Align.Applications import MafftCommandline
|
| 22 |
+
from Bio import Phylo
|
| 23 |
+
from Bio.Phylo.TreeConstruction import DistanceCalculator, DistanceTreeConstructor
|
| 24 |
+
import io
|
| 25 |
+
|
| 26 |
+
os.environ['GRADIO_SERVER_NAME'] = '0.0.0.0'
|
| 27 |
+
title='DNAptaESM2 Model Infernence'
|
| 28 |
+
desc='AptaBLE (cross-attention network), trained to predict the likelihood a DNA aptamer will form a complex with a target protein!\n\nPass in a FASTA-formatted file of all aptamers and input your protein target amino acid sequence. Your output scores are available for download via an Excel file.'
|
| 29 |
+
|
| 30 |
+
global pipeline
|
| 31 |
+
|
| 32 |
+
pipeline = AptaTransPipeline_Dist(
|
| 33 |
+
lr=1e-6,
|
| 34 |
+
weight_decay=None,
|
| 35 |
+
epochs=None,
|
| 36 |
+
model_type=None,
|
| 37 |
+
model_version=None,
|
| 38 |
+
model_save_path=None,
|
| 39 |
+
accelerate_save_path=None,
|
| 40 |
+
tensorboard_logdir=None,
|
| 41 |
+
d_model=128,
|
| 42 |
+
d_ff=512,
|
| 43 |
+
n_layers=6,
|
| 44 |
+
n_heads=8,
|
| 45 |
+
dropout=0.1,
|
| 46 |
+
load_best_pt=True, # already loads the pretrained model using the datasets included in repo -- no need to run the bottom two cells
|
| 47 |
+
device='cuda',
|
| 48 |
+
seed=1004)
|
| 49 |
+
|
| 50 |
+
def comparison(protein, aptamer_file, analysis):
|
| 51 |
+
print('analysis: ', analysis)
|
| 52 |
+
display = []
|
| 53 |
+
table_data = pd.DataFrame()
|
| 54 |
+
r_names, aptamers = read_fasta(aptamer_file)
|
| 55 |
+
proteins = [protein for i in range(len(aptamers))]
|
| 56 |
+
df = pd.DataFrame(columns=['Protein', 'Protein Seq', 'Aptamer', 'Aptamer Seq', 'Score'])
|
| 57 |
+
# print('Number of aptamers: ', len(aptamers))
|
| 58 |
+
scores = get_scores(aptamers, proteins)
|
| 59 |
+
df['Protein'] = ['protein_prov.']*len(aptamers)
|
| 60 |
+
df['Aptamer'] = r_names
|
| 61 |
+
df['Protein Seq'] = proteins
|
| 62 |
+
df['Aptamer Seq'] = aptamers
|
| 63 |
+
df['Score'] = scores
|
| 64 |
+
|
| 65 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".xlsx") as temp_file:
|
| 66 |
+
with pd.ExcelWriter(temp_file.name, engine='openpyxl') as writer:
|
| 67 |
+
df.to_excel(writer, index=False)
|
| 68 |
+
temp_file_path = temp_file.name
|
| 69 |
+
|
| 70 |
+
print('Saving to excel!')
|
| 71 |
+
df.to_excel(f'{aptamer_file}.xlsx')
|
| 72 |
+
|
| 73 |
+
torch.cuda.empty_cache()
|
| 74 |
+
|
| 75 |
+
return '\n'.join(display), temp_file_path
|
| 76 |
+
|
| 77 |
+
def read_fasta(file_path):
|
| 78 |
+
headers = []
|
| 79 |
+
sequences = []
|
| 80 |
+
with open(file_path, 'r') as file:
|
| 81 |
+
content = file.readlines()
|
| 82 |
+
for i in range(0, len(content), 2):
|
| 83 |
+
header = content[i].strip()
|
| 84 |
+
if header.startswith('>'):
|
| 85 |
+
headers.append(header)
|
| 86 |
+
sequences.append(content[i+1].strip())
|
| 87 |
+
return headers, sequences
|
| 88 |
+
|
| 89 |
+
def get_scores(aptamers, proteins):
|
| 90 |
+
pipeline.model.to('cuda')
|
| 91 |
+
scores = pipeline.inference(aptamers, proteins, [0]*len(aptamers))
|
| 92 |
+
pipeline.model.to('cpu')
|
| 93 |
+
return scores
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
iface = gr.Interface(
|
| 97 |
+
fn=comparison,
|
| 98 |
+
inputs=[
|
| 99 |
+
gr.Textbox(lines=2, placeholder="Protein"),
|
| 100 |
+
gr.File(type="filepath"),
|
| 101 |
+
],
|
| 102 |
+
outputs=[
|
| 103 |
+
gr.Textbox(placeholder="Scores"),
|
| 104 |
+
gr.File(label="Download Excel")
|
| 105 |
+
],
|
| 106 |
+
description=desc
|
| 107 |
+
)
|
| 108 |
+
|
| 109 |
+
iface.launch()
|