app.py

import streamlit as st
import pandas as pd
from alignment.needleman_wunsch import needleman_wunsch
from alignment.smith_waterman import smith_waterman
from alignment.dot_matrix import plot_dot_matrix
from alignment.word_method import word_alignment
from Bio import SeqIO
import io
import os
import re

def clean_dna_sequence(seq):
    cleaned = re.sub(r'[^ATGC]', '', seq.upper())
    return cleaned

st.set_page_config(page_title="GeneAligner", layout="wide")
st.title("🧬 GeneAligner: Bioinformatics Sequence Alignment Tool")

uploaded_file1 = st.file_uploader("Upload Sequence A (FASTA, GenBank, TXT)", type=["fasta", "fa", "gb", "txt"])
uploaded_file2 = st.file_uploader("Upload Sequence B (FASTA, GenBank, TXT)", type=["fasta", "fa", "gb", "txt"])

def read_sequence(uploaded_file):
    if uploaded_file:
        content = uploaded_file.read().decode("utf-8")
        file_type = uploaded_file.name.split(".")[-1].lower()
        if file_type in ["fa", "fasta"]:
            return "".join([line for line in content.splitlines() if not line.startswith(">")])
        elif file_type == "gb":
            try:
                record = SeqIO.read(io.StringIO(content), "genbank")
                return str(record.seq)
            except:
                return ""
        elif file_type == "txt":
            return content.replace("\n", "").strip()
    return ""

# Alignment match line generator
def get_match_line(seq1, seq2):
    line = ""
    for a, b in zip(seq1, seq2):
        if a == b:
            line += "|"
        elif a == "-" or b == "-":
            line += " "
        else:
            line += "."
    return line

# Color-coded HTML alignment
def render_alignment_with_color(seq1, seq2):
    html_seq1, html_seq2 = "", ""
    for a, b in zip(seq1, seq2):
        if a == b:
            html_seq1 += f'<span style="color:green">{a}</span>'
            html_seq2 += f'<span style="color:green">{b}</span>'
        elif a == "-" or b == "-":
            html_seq1 += f'<span style="color:gray">{a}</span>'
            html_seq2 += f'<span style="color:gray">{b}</span>'
        else:
            html_seq1 += f'<span style="color:red">{a}</span>'
            html_seq2 += f'<span style="color:red">{b}</span>'
    st.markdown("#### 🔬 Aligned Sequences (Colored)", unsafe_allow_html=True)
    st.markdown(f"<code>{html_seq1}</code>", unsafe_allow_html=True)
    st.markdown(f"<code>{html_seq2}</code>", unsafe_allow_html=True)

from groq import Groq

# Initialize Groq Client
client = Groq(api_key=os.getenv("GROQ_API_KEY"))

# Inputs
original_seq1 = read_sequence(uploaded_file1) or st.text_area("Or paste Sequence A", height=150)
original_seq2 = read_sequence(uploaded_file2) or st.text_area("Or paste Sequence B", height=150)
seq1 = clean_dna_sequence(original_seq1)
seq2 = clean_dna_sequence(original_seq2)

# Warn if cleaning was needed
if original_seq1 != seq1:
    st.warning("⚠️ Invalid characters removed from Sequence A. Only A, T, G, and C are allowed.")

if original_seq2 != seq2:
    st.warning("⚠️ Invalid characters removed from Sequence B. Only A, T, G, and C are allowed.")
method = st.selectbox("🔧 Choose Alignment Method", 
                      ["Dot Matrix", "Needleman-Wunsch", "Smith-Waterman", "Word Method"])

if st.button("🔍 Align Sequences"):
    if not seq1 or not seq2:
        st.warning("Please input both sequences.")
    else:
        align1 = align2 = ""
        if method == "Dot Matrix":
            plot_dot_matrix(seq1, seq2)
            align1, align2, score, identity = seq1, seq2, "N/A", "N/A"
            
    #else: 
     #   align1 = align2 = ""
        if method == "Needleman-Wunsch":
            score, align1, align2, matrix,match_line,identity = needleman_wunsch(seq1, seq2)
            st.write(f"**Global Alignment Score:** {score}")
            # Show Scoring Matrix
            st.markdown("#### 🧮 Alignment Scoring Matrix")
            row_labels = [f"{i}-{char}" for i, char in enumerate("-" + seq1)]
            col_labels = [f"{i}-{char}" for i, char in enumerate("-" + seq2)]
            df_matrix = pd.DataFrame(matrix, index=row_labels, columns=col_labels)
            st.dataframe(df_matrix.style.background_gradient(cmap='Blues'))
          
        elif method == "Smith-Waterman": 
            score, align1, align2,matrix,match_line,identity = smith_waterman(seq1, seq2)
            st.write(f"**Local Alignment Score:** {score}")
            #Scoring matrix
            st.markdown("#### 🧮 Alignment Scoring Matrix")
            row_labels = [f"{i}-{char}" for i, char in enumerate("-" + seq1)]
            col_labels = [f"{i}-{char}" for i, char in enumerate("-" + seq2)]
            df_matrix = pd.DataFrame(matrix, index=row_labels, columns=col_labels)
            st.dataframe(df_matrix.style.background_gradient(cmap='Blues'))
           # if align1 and align2:
            #    with st.expander("💬 Ask the bot to explain your alignment result"):
             #       explanation = interpret_alignment(method, score, identity, align1, align2)
              #      st.markdown(explanation)
            #else:
             #   st.warning("No alignment found to interpret.")
        elif method == "Word Method":
            word_alignment(seq1, seq2, word_size=3)
            score,Identity="N/A","N/A"
                  
                

        # Output alignment if available
        if align1 and align2:
            st.markdown("#### 🧬 Aligned Sequences")
            st.code(align1)
            st.code(get_match_line(align1, align2))
            st.code(align2)

            render_alignment_with_color(align1, align2)

            # Download as CSV
            a_list = list(align1)
            b_list = list(align2)

            max_len = max(len(a_list), len(b_list))
            a_list += [""] * (max_len - len(a_list))  # pad with empty strings
            b_list += [""] * (max_len - len(b_list))

            df = pd.DataFrame({"Sequence A": a_list, "Sequence B": b_list})
            csv = df.to_csv(index=False).encode("utf-8")
            st.download_button("📥 Download Alignment as CSV", csv, "alignment_result.csv", "text/csv")
        

st.sidebar.header("💬 Bioinformatics Chatbot")
user_question = st.sidebar.text_area("Ask me anything about alignments")
if st.sidebar.button("Send"):
    if user_question.strip():
        response = client.chat.completions.create(
            model="llama-3.1-8b-instant",
            messages=[{"role": "user", "content": user_question}],
        )
        st.sidebar.write("**Bot:**", response.choices[0].message.content)