ASCARIS.py

import streamlit as st
import pandas as pd 
from os import path
import sys
import streamlit.components.v1 as components
sys.path.append('code/')
#sys.path.append('ASCARIS/code/') 
import pdb_featureVector
import alphafold_featureVector
import argparse
from st_aggrid import AgGrid, GridOptionsBuilder, JsCode,GridUpdateMode
import base64
showWarningOnDirectExecution = False


from datasets import Dataset, concatenate_datasets
MAX_SAMPLES_IN_MEMORY = 1000
samples_in_dset = 0
dset = Dataset.from_dict({"col1": [], "col2": []})  # empty dataset
path_to_save_dir = "HUBioData/input_files"
num_chunks = 0
for example_dict in custom_example_dict_streamer("HUBioData/AlphafoldStructures"):
    dset = dset.add_item(example_dict)
    samples_in_dset += 1
    if samples_in_dset == MAX_SAMPLES_IN_MEMORY:
        samples_in_dset = 0
        dset.save_to_disk(f"{path_to_save_dir}{num_chunks}")
        num_chunks =+ 1
        dset = Dataset.from_dict({"col1": [], "col2": []})  # empty dataset
if samples_in_dset > 0:
    dset.save_to_disk(f"{path_to_save_dir}{num_chunks}")
    num_chunks =+ 1
loaded_dsets = []  # memory-mapped
for chunk_num in range(num_chunks):
    dset = Dataset.load_from_disk(f"{path_to_save_dir}{chunk_num}") 
    loaded_dsets.append(dset)
final_dset = concatenate_datasets(dset)
st.write('FİNAL DSET')
st.write(final_dset)





def convert_df(df):
   return df.to_csv(index=False).encode('utf-8')

    
# Check if 'key' already exists in session_state
# If not, then initialize it
if 'visibility' not in st.session_state:
    st.session_state['visibility'] = 'visible'
    st.session_state.disabled =  False

original_title = '<p style="font-family:Trebuchet MS; color:#FD7456; font-size: 25px; font-weight:bold; text-align:center">ASCARIS</p>'
st.markdown(original_title, unsafe_allow_html=True)
original_title = '<p style="font-family:Trebuchet MS; color:#FD7456; font-size: 25px; font-weight:bold; text-align:center">(Annotation and StruCture-bAsed RepresentatIon of Single amino acid variations)</p>'
st.markdown(original_title, unsafe_allow_html=True)
 
st.write('')
st.write('')
st.write('')
st.write('')

with st.form('mform', clear_on_submit=False):
    source = st.selectbox('Select the protein structure resource (1: PDB-SwissModel-Modbase, 2: AlphaFold)',[1,2])
    ###source = 1
    impute = st.selectbox('Imputation',[True, False])
    input_data = st.text_input('Enter SAV data points (Example: Q00889-H-85-D, or Q00889-H-85-D,Q16363-Y-498-H)')
            


    parser = argparse.ArgumentParser(description='ASCARIS')
    
    #parser.add_argument('-s', '--source_option',
    #                    help='Selection of input structure data.\n 1: PDB Structures (default), 2: AlphaFold Structures',
    #                    default=1)
    #parser.add_argument('-i', '--input_datapoint',
    #                    help='Input file or query datapoint\n Option 1: Comma-separated list of identifiers (UniProt ID-wt residue-position-mutated residue (e.g. Q9Y4W6-N-432-T or Q9Y4W6-N-432-T, Q9Y4W6-N-432-T)) \n Option 2: Enter comma-separated file path')
    #
    #parser.add_argument('-impute', '--imputation_state', default='True',
    #                    help='Whether resulting feature vector should be imputed or not. Default True.')
    
    #args = parser.parse_args()
    
    input_set = input_data
    ###mode = 1
    impute = impute
    submitted = st.form_submit_button(label="Submit", help=None, on_click=None, args=None, kwargs=None, type="secondary", disabled=False, use_container_width=False)
    print('*****************************************')
    print('Feature vector generation is in progress. \nPlease check log file for updates..')
    print('*****************************************')
    ###mode = int(mode)
    mode = source
    
selected_df = pd.DataFrame()
st.write('The online tool may be slow, especially while processing multiple SAVs, please consider using the local programmatic version at https://github.com/HUBioDataLab/ASCARIS/')
if submitted:
    with st.spinner('In progress...This may take a while...'):
        try:
            if mode == 1:
                selected_df = pdb_featureVector.pdb(input_set, mode, impute)    
                
            elif mode == 2:
                selected_df = alphafold_featureVector.alphafold(input_set, mode, impute)
            else:
                selected_df =  pd.DataFrame()

        except:
            selected_df = pd.DataFrame()
            pass

    if selected_df is None:
        st.success('Feature vector failed. Check log file.')

    else:
        if len(selected_df) != 0 :
            st.write(selected_df)
            st.success('Feature vector successfully created.')
            csv = convert_df(selected_df)
    
            st.download_button("Press to Download the Feature Vector", csv,f"ASCARIS_SAV_rep_{input_set}.csv","text/csv",key='download-csv')

        else:
            st.success('Feature vector failed. Check log file.')