# app.py

import gradio as gr
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer

def generate_sequences(model_name, prompt):
    if model_name == "nferruz/ProtGPT2":
        protgpt2 = pipeline('text-generation', model="nferruz/ProtGPT2")
        sequences = protgpt2(prompt, max_length=100, do_sample=True, top_k=950, repetition_penalty=1.2, num_return_sequences=10, eos_token_id=0)
        return "\n".join([seq['generated_text'] for seq in sequences])
    elif model_name == "lightonai/RITA_xl":
        model = AutoModelForCausalLM.from_pretrained("lightonai/RITA_xl", trust_remote_code=True)
        tokenizer = AutoTokenizer.from_pretrained("lightonai/RITA_xl")
        rita_gen = pipeline('text-generation', model=model, tokenizer=tokenizer)
        sequences = rita_gen(prompt, max_length=20, do_sample=True, top_k=950, repetition_penalty=1.2, num_return_sequences=2, eos_token_id=2)
        return "\n".join([seq['generated_text'].replace(' ', '') for seq in sequences])
    else:
        return "Model not supported"

model_options = ["nferruz/ProtGPT2", "lightonai/RITA_xl"]

gr.Interface(
    fn=generate_sequences,
    inputs=[
        gr.Dropdown(model_options, label="Select Model"),
        gr.Textbox(lines=2, placeholder="Enter your prompt here...", label="Prompt")
    ],
    outputs="text",
    title="Novel Protein Sequence Generation",
    description="Generate sequences using selected protein language models."
).launch()