| import numpy as np |
| import pandas as pd |
| import plotly.express as px |
| import streamlit as st |
|
|
|
|
| def saturated_mutations(dna_seq): |
| dna_seq = dna_seq.upper().replace("U", "T") |
| dna_seq = "".join(c for c in dna_seq if c in "ATCG") |
| n = len(dna_seq) |
| bases = np.array(list("ATCG")) |
| seq_array = np.array(list(dna_seq)) |
| replace_matrix = np.array([bases[bases != base] for base in seq_array]) |
| result = np.tile(seq_array, (3 * n, 1)) |
| rows = np.repeat(np.arange(n), 3) |
| result[np.arange(3 * n), rows] = replace_matrix.flatten() |
| result = np.vstack([seq_array, result]) |
| return ["".join(row) for row in result] |
|
|
|
|
| if __name__ == "__main__": |
| model = st.session_state["model"] |
| input_format = st.session_state["input_format"] |
| predict = st.session_state["predict"] |
| mrna_env = st.session_state["mrna_env"] |
| UTR3_GLUC = st.session_state["UTR3_GLUC"] |
| CDS_GLUC = st.session_state["CDS_GLUC"] |
| st.set_page_config(layout="wide") |
| st.title("UTR-CODE") |
| with st.form("single_input_form"): |
| utr5_input = st.text_area( |
| "5' UTR:", |
| key="utr5_input", |
| max_chars=1000, |
| help="Only 'A', 'C', 'G', 'T' and 'U' is allowed, other characters will be ignored", |
| ) |
| with st.expander("Optional Input"): |
| optional_input_col1, optional_input_col2 = st.columns(2) |
| with optional_input_col1: |
| cds_input = st.text_area( |
| "CDS", |
| value=CDS_GLUC, |
| key="cds_input", |
| max_chars=4500, |
| help="GLUC's CDS is used by default.", |
| ) |
| with optional_input_col2: |
| utr3_input = st.text_area( |
| "3' UTR:", |
| value=UTR3_GLUC, |
| key="utr3_input", |
| max_chars=4000, |
| help="GLUC's UTR3 is used by default.", |
| ) |
| mutation_rounds = st.slider( |
| "ISM Rounds", |
| min_value=1, |
| max_value=10, |
| value=5, |
| step=1, |
| label_visibility="visible", |
| ) |
| submitted = st.form_submit_button("Submit", use_container_width=True) |
| if submitted: |
| if not "".join([c for c in utr5_input.upper() if c in "ATCGU"]): |
| st.warning("Please input 5' UTR") |
| elif any(c not in "ATCGU\n\t, " for c in f"{utr5_input.upper()}{cds_input.upper()}{utr3_input.upper()}"): |
| st.warning("Only 'A', 'C', 'G', 'T' and 'U' is allowed.") |
| else: |
| utr5_optimize_use = "".join( |
| [c for c in utr5_input.upper().replace("U", "T") if c in "ATCG"] |
| ) |
| results = dict() |
| utr, cds = input_format(utr5_optimize_use, cds_input, utr3_input) |
| results[0] = { |
| "rpf": (np.expm1(predict(model, utr, cds, mrna_env)[0]) / 5) / 108.7144 , |
| "seq": utr5_optimize_use, |
| "base_change": "Origin", |
| } |
| cds_seq = [ |
| cds_input, |
| ] * (3 * len(utr5_input) + 1) |
| utr3_seq = [ |
| utr3_input, |
| ] * (3 * len(utr5_input) + 1) |
| progress_bar = st.empty() |
| progress_bar = st.progress(0, f"Progress: 0%") |
| progress_text = st.empty() |
| chart_placeholder = st.empty() |
| for i in range(1, mutation_rounds + 1): |
| used_seq = saturated_mutations(utr5_optimize_use) |
| utr, cds = input_format(used_seq, cds_seq, utr3_seq) |
| result = predict(model, utr, cds, mrna_env) |
| best_idx = np.argmax(result) |
| max_val = result[best_idx] |
| base_change = next( |
| ( |
| f"{i+1}:{c1}->{c2}" |
| for i, (c1, c2) in enumerate( |
| zip(utr5_optimize_use, used_seq[best_idx]) |
| ) |
| if c1 != c2 |
| ), |
| "No change", |
| ) |
| utr5_optimize_use = used_seq[best_idx] |
| results[i] = { |
| "rpf": (np.expm1(max_val) / 5) / 108.7144, |
| "seq": utr5_optimize_use, |
| "base_change": base_change, |
| } |
| progress_bar.progress( |
| i / mutation_rounds, |
| f"Progress: {i / mutation_rounds * 100:.0f}%", |
| ) |
| df = pd.DataFrame( |
| [ |
| { |
| "Round": k, |
| "Predicted Value": float(v["rpf"]), |
| "UTR5 Sequence": v["seq"], |
| "Base Change": v["base_change"], |
| } |
| for k, v in results.items() |
| ] |
| ) |
| fig = px.line( |
| df, |
| x="Round", |
| y="Predicted Value", |
| text="Base Change", |
| markers=True, |
| ) |
| fig.update_layout( |
| template="plotly_white", |
| title="UTR5 Optimization Progress", |
| title_font_family="Arial, sans-serif", |
| title_font_color="black", |
| xaxis_title="Mutation Round", |
| yaxis_title="Predicted Value (RPF)", |
| xaxis_title_font=dict(color="black"), |
| yaxis_title_font=dict(color="black"), |
| font_family="Arial, sans-serif", |
| font_color="black", |
| margin=dict(l=50, r=50, t=80, b=50), |
| ) |
| fig.update_xaxes( |
| showgrid=False, |
| showline=True, |
| linecolor="black", |
| tickcolor="black", |
| ticks="outside", |
| tickfont_color="black", |
| ) |
| fig.update_yaxes( |
| showgrid=False, |
| showline=True, |
| linecolor="black", |
| tickcolor="black", |
| ticks="outside", |
| tickfont_color="black", |
| tickformat=".2f", |
| ) |
| fig.update_traces( |
| textposition="top center", |
| marker=dict(color="#cf3326", line=dict(color="#cf3326")), |
| line=dict(color="#cf3326"), |
| textfont_color="black", |
| ) |
| fig.update_xaxes(dtick=1) |
| chart_placeholder.plotly_chart(fig, use_container_width=True) |
| progress_bar.empty() |
| st.download_button( |
| "Download Result", |
| data=df.to_csv(index=False), |
| file_name="results.csv", |
| mime="text/csv", |
| use_container_width=True, |
| icon="📥", |
| ) |
|
|
