app.py

import os
import gradio as gr
import pandas as pd
from optimizer import optimization

BASE_DIR = os.path.dirname(__file__)
SUMMARY_PATH = os.path.join(BASE_DIR, "region_sweep_summary.csv")

AA_ORDER = list("ACDEFGHIKLMNPQRSTVWY*")
AA_ALLOWED = set(AA_ORDER)

def _clean_aa_seq(raw: str) -> str:
    """
    Keep only allowed amino-acid symbols defined in AA_ORDER.
    Removes spaces, punctuation, digits, and any other letters.
    Example: ' mkk ll;ptaA ' -> 'MKKLLPTAA'
    """
    s = (raw or "").upper()
    return "".join(ch for ch in s if ch in AA_ALLOWED)

def aa_percent_to_onecol_df(aa_percent: dict, digits: int = 0) -> pd.DataFrame:
    rows = []
    order = AA_ORDER if set(aa_percent).issubset(set(AA_ORDER)) else sorted(aa_percent)
    for aa in order:
        mix = aa_percent.get(aa, {})
        if not mix:
            rows.append([aa, "—"])
            continue
        parts = sorted(mix.items(), key=lambda kv: (-kv[1], kv[0]))
        cell = " - ".join(f"{cod} ({val*100:.{digits}f}%)" for cod, val in parts)
        rows.append([aa, cell])
    return pd.DataFrame(rows, columns=["AA", "Codon percentage"])

# NOTE: run() params MUST match the number/order of Interface inputs
def run(aa_seq: str, use_wobble: bool):
    cleaned = _clean_aa_seq(aa_seq)
    if not cleaned:
        raise gr.Error("Input sequence contains no valid amino-acid characters after cleaning.")

    # pass the checkbox value straight through
    designed_nt, aa_percent, gc_percent, _ = optimization(
        summary_path=SUMMARY_PATH,
        aa_seq=cleaned,
        use_wobble=use_wobble,     # ← your current optimizer flag name
        wobble_scale=1.0,            # adjust if you expose a slider later
        use_percent_intervals=True,
    )

    aa_table = aa_percent_to_onecol_df(aa_percent, digits=0)
    if not isinstance(gc_percent, pd.DataFrame):
        gc_percent = pd.DataFrame(gc_percent)

    return designed_nt, aa_table, gc_percent

# ---- Gradio Interface ----
iface = gr.Interface(
    fn=run,
    inputs=[
        gr.Textbox(label="Amino Acid Sequence", lines=5, placeholder="e.g. MKKLLPTAA..."),
        gr.Checkbox(
            label="Use wobble score",
            value=True,  # default ON
            info="When ON, adds a 3rd-base preference (e.g., C favored; A/T slightly penalized) during seeding & filling. Turn OFF to rely only on k-mer motifs."
        ),
    ],
    outputs=[
        gr.Textbox(label="Optimized Nucleotide Sequence"),
        gr.Dataframe(label="Codon Usage Percent (per AA)", wrap=True),
        gr.Dataframe(label="GC Content (%)", wrap=True),
    ],
    title="Codon Optimizer",
    description="Toggle wobble to include/exclude position-3 base scoring in the design.",
    flagging_mode="never",
)

if __name__ == "__main__":
    # iface.queue().launch()
    iface.launch()