update

app.py

@@ -1,64 +1,73 @@
+import os
+import re
+import torch
 import gradio as gr
-from huggingface_hub import InferenceClient
+import numpy as np
+import sklearn
+from tqdm import tqdm
+from datasets import load_dataset, DatasetDict
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# Automatically detect GPU or use CPU
+device = "cuda" if torch.cuda.is_available() else "cpu"
 
+# Default model path
+model_tokenizer_path = "zehui127/Omni-DNA-Multitask"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# Load tokenizer and model with trusted remote code
+tokenizer = AutoTokenizer.from_pretrained(model_tokenizer_path, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(model_tokenizer_path, trust_remote_code=True).to(device)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+# List of available tasks
+tasks = ['H3', 'H4', 'H3K9ac', 'H3K14ac', 'H4ac', 'H3K4me1', 'H3K4me2', 'H3K4me3', 'H3K36me3', 'H3K79me3']
 
-    messages.append({"role": "user", "content": message})
+def preprocess_response(response, mask_token="[MASK]"):
+    """Extracts the response after the [MASK] token."""
+    if mask_token in response:
+        response = response.split(mask_token, 1)[1]
+    response = re.sub(r'^[\sATGC]+', '', response)
+    return response
 
-    response = ""
+def generate(dna_sequence, task_type, sample_num=1):
+    """
+    Generates a response based on the DNA sequence and selected task.
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    Args:
+        dna_sequence (str): The input DNA sequence.
+        task_type (str): The selected task type.
+        sample_num (int): Number of samples for the generation process.
 
-        response += token
-        yield response
+    Returns:
+        str: Predicted function label.
+    """
+    dna_sequence = dna_sequence + task_type +"[MASK]"
+    tokenized_message = tokenizer(
+        [dna_sequence], return_tensors='pt', return_token_type_ids=False, add_special_tokens=True
+    ).to(device)
 
+    response = model.generate(**tokenized_message, max_new_tokens=sample_num, do_sample=False)
+    reply = tokenizer.batch_decode(response, skip_special_tokens=False)[0].replace(" ", "")
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+    return extract_label(reply, task_type)
+
+def extract_label(message, task_type):
+    """Extracts the prediction label from the model's response."""
+    task_type = '[MASK]'
+    answer = message.split(task_type)[1]
+    match = re.search(r'\d+', answer)
+    return match.group() if match else "No valid prediction"
+
+# Gradio interface
+interface = gr.Interface(
+    fn=generate,
+    inputs=[
+        gr.Textbox(label="Input DNA Sequence", placeholder="Enter a DNA sequence"),
+        gr.Dropdown(choices=tasks, label="Select Task Type"),
     ],
+    outputs=gr.Textbox(label="Predicted Function"),
+    title="Omni-DNA Multitask Prediction",
+    description="Select a DNA-related task and input a sequence to generate function predictions.",
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    interface.launch()

requirements.txt

@@ -1 +1,6 @@
-huggingface_hub==0.25.2
+huggingface_hub==0.25.2
+torch
+transformers
+gradio
+datasets
+ai2-olmo