通过VSCode连接Bohrium节点开发机制作镜像 App

原创

用户10497140

修改于 2025-04-08 10:08:56

2700

App、镜像的对外发布，总是需要一个“可用的版本”，而可用版本的运行，依托的是开发者创建好的镜像。

关键镜像的制作和封装，熟悉Docker操作的同学，自无需多言。

本位旨在为不熟悉Docker操作的开发者用户，提供一个便捷方法，完成镜像封装。

在 Bohrium 节点管理页右上角选择“创建容器节点”，本案例中镜像选择 ubuntu:20.04-py3.10，所属项目选择节点对应项目，机器类型、系统盘以及自动关机选项无需修改，保持默认值即可。创建后通常需要 1 分钟左右的时间来启动管理节点，等节点状态从“启动中”变为“运行中”时，即可登录使用。

SSH 登录创建的 Bohrium Node，在节点中导入代码，并安装好代码运行所需的环境依赖。

注意，代码和环境依赖，不能放在 /personal 、/bohr 或 /share 目录下，这几个目录下的文件，并不会随项目被打包到镜像当中。

Bohrium 提供了网页版 SSH 工具 Web Shell，也支持您通过本地终端登录节点。

也推荐使用本地的VSCode，连接Bohrium节点使用。可参考如何用本地VSCode连接Bohrium节点

此外，开发机的端口50001～50005对外可用，可以用来提供对外服务。

打开本地VSCode SSH 工具链接，github找到源码

下载源码，创建环境和依赖

解决conda init 先运行conda init 后另起终端 conda activate apt

touch /root/AptaTrans/requirements.txt
pip install -r /root/AptaTrans/requirements.txt

torch==2.0.1
tqdm==4.62.1
numpy==1.24.3
scikit-learn==1.2.2
sqlite3
pickle
matplotlib
einops

运行示例examples.ipynb 内核选择刚刚创建的apt 环境

frontend/apt_gradio.py

import gradio as gr
import os
import sys
from pathlib import Path

# Add the project root directory to the Python path
project_root = str(Path(__file__).parent.parent)
sys.path.append(project_root)

from aptatrans_pipeline import AptaTransPipeline
import torch

def predict_api_score(aptamer, target):
    try:
        # Initialize pipeline with correct data path
        pipeline = AptaTransPipeline(
            dim=128,
            mult_ff=2,
            n_layers=6,
            n_heads=8,
            dropout=0.1,
            load_best_pt=False,
            load_best_model=True,
            save_name='default',
            device='cuda:0' if torch.cuda.is_available() else 'cpu',
            seed=1004,
            data_dir=os.path.join(project_root, 'data')
        )
        
        # Make prediction
        score = pipeline.inference(aptamer, target)
        return f"Predicted API Score: {score[0][0]:.4f}"
    except Exception as e:
        return f"Error during prediction: {str(e)}"

def recommend_aptamers(target):
    try:
        # Initialize pipeline with correct data path
        pipeline = AptaTransPipeline(
            dim=128,
            mult_ff=2,
            n_layers=6,
            n_heads=8,
            dropout=0.1,
            load_best_pt=False,
            load_best_model=True,
            save_name='default',
            device='cuda:0' if torch.cuda.is_available() else 'cpu',
            seed=1004,
            data_dir=os.path.join(project_root, 'data')
        )
        
        # Get recommendations
        results = pipeline.recommend(target, n_aptamers=5, depth=40, iteration=1000)
        
        # Format output
        output = "Recommended Aptamers:\n\n"
        for idx, result in results.items():
            output += f"Candidate {idx + 1}:\n"
            output += f"Sequence: {result['candidate']}\n"
            output += f"Score: {result['score'].item():.4f}\n\n"
        
        return output
    except Exception as e:
        return f"Error during recommendation: {str(e)}"

# Create Gradio interface
with gr.Blocks(title="AptaTrans") as demo:
    gr.Markdown("# AptaTrans - Aptamer-Protein Interaction Prediction")
    
    with gr.Tab("Predict API Score"):
        gr.Markdown("## Predict Aptamer-Protein Interaction Score")
        with gr.Row():
            with gr.Column():
                aptamer_input = gr.Textbox(label="Aptamer Sequence", placeholder="Enter aptamer sequence...")
                target_input = gr.Textbox(label="Target Protein Sequence", placeholder="Enter target protein sequence...")
                predict_btn = gr.Button("Predict Score")
            with gr.Column():
                output_score = gr.Textbox(label="Prediction Result")
        
        predict_btn.click(
            fn=predict_api_score,
            inputs=[aptamer_input, target_input],
            outputs=output_score
        )
    
    with gr.Tab("Recommend Aptamers"):
        gr.Markdown("## Recommend Candidate Aptamers")
        with gr.Row():
            with gr.Column():
                target_input_recommend = gr.Textbox(label="Target Protein Sequence", placeholder="Enter target protein sequence...")
                recommend_btn = gr.Button("Recommend Aptamers")
            with gr.Column():
                output_recommend = gr.TextArea(
                    label="Recommendation Results",
                    lines=15,  # Set number of visible lines
                    max_lines=50  # Set maximum number of lines
                )
        
        recommend_btn.click(
            fn=recommend_aptamers,
            inputs=target_input_recommend,
            outputs=output_recommend
        )

if __name__ == "__main__":
    demo.launch(share=True)

试运行

CTGATTTTCCTTCCAGGCACCAC

MAVEGGMKCVKFLLYVLLLAFCACAVGLIAVGVGAQLVLSQTIIQGATPGSLLPVVIIAVGVFLFLVAFVGCCGACKENYCLMITFAIFLSLIMLVEVAAAIAGYVFRDKVMSEFNNNFRQQMENYPKNNHTASILDRMQADFKCCGAANYTDWEKIPSMSKNRVPDSCCINVTVGCGINFNEKAIHKEGCVEKIGGWLRKNVLVVAAAALGIAFVEVLGIVFACCLVKSIRSGYEVM

另起终端 nvidia-smi 查看显存使用

app.py Fastapi封装

import os
import sys
os.system("pip install -r /root/AptaTrans/requirements.txt")
from pathlib import Path
import gradio as gr
import torch
from fastapi import FastAPI
import uvicorn

# Add the project root directory to the Python path
project_root = str(Path(__file__).parent)
sys.path.append(project_root)

from aptatrans_pipeline import AptaTransPipeline

def predict_api_score(aptamer, target):
    try:
        # Initialize pipeline with correct data path
        pipeline = AptaTransPipeline(
            dim=128,
            mult_ff=2,
            n_layers=6,
            n_heads=8,
            dropout=0.1,
            load_best_pt=False,
            load_best_model=True,
            save_name='default',
            device='cuda:0' if torch.cuda.is_available() else 'cpu',
            seed=1004,
            data_dir=os.path.join(project_root, 'data')
        )
        
        # Make prediction
        score = pipeline.inference(aptamer, target)
        return f"Predicted API Score: {score[0][0]:.4f}"
    except Exception as e:
        return f"Error during prediction: {str(e)}"

def recommend_aptamers(target):
    try:
        # Initialize pipeline with correct data path
        pipeline = AptaTransPipeline(
            dim=128,
            mult_ff=2,
            n_layers=6,
            n_heads=8,
            dropout=0.1,
            load_best_pt=False,
            load_best_model=True,
            save_name='default',
            device='cuda:0' if torch.cuda.is_available() else 'cpu',
            seed=1004,
            data_dir=os.path.join(project_root, 'data')
        )
        
        # Get recommendations
        results = pipeline.recommend(target, n_aptamers=5, depth=40, iteration=1000)
        
        # Format output
        output = "Recommended Aptamers:\n\n"
        for idx, result in results.items():
            output += f"Candidate {idx + 1}:\n"
            output += f"Sequence: {result['candidate']}\n"
            output += f"Score: {result['score'].item():.4f}\n\n"
        
        return output
    except Exception as e:
        return f"Error during recommendation: {str(e)}"

# Create Gradio interface
with gr.Blocks(title="AptaTrans") as demo:
    gr.Markdown("# AptaTrans - Aptamer-Protein Interaction Prediction")
    
    with gr.Tab("Predict API Score"):
        gr.Markdown("## Predict Aptamer-Protein Interaction Score")
        with gr.Row():
            with gr.Column():
                aptamer_input = gr.Textbox(label="Aptamer Sequence", placeholder="Enter aptamer sequence...")
                target_input = gr.Textbox(label="Target Protein Sequence", placeholder="Enter target protein sequence...")
                predict_btn = gr.Button("Predict Score")
            with gr.Column():
                output_score = gr.Textbox(label="Prediction Result")
        
        predict_btn.click(
            fn=predict_api_score,
            inputs=[aptamer_input, target_input],
            outputs=output_score
        )
    
    with gr.Tab("Recommend Aptamers"):
        gr.Markdown("## Recommend Candidate Aptamers")
        with gr.Row():
            with gr.Column():
                target_input_recommend = gr.Textbox(label="Target Protein Sequence", placeholder="Enter target protein sequence...")
                recommend_btn = gr.Button("Recommend Aptamers")
            with gr.Column():
                output_recommend = gr.TextArea(
                    label="Recommendation Results",
                    lines=15,
                    max_lines=50
                )
        
        recommend_btn.click(
            fn=recommend_aptamers,
            inputs=target_input_recommend,
            outputs=output_recommend
        )

# Create FastAPI app
app = FastAPI(title="AptaTrans API")

# Mount Gradio app to FastAPI
app = gr.mount_gradio_app(app, demo, path="/")

if __name__ == "__main__":
    # Run the application on port 50001
    uvicorn.run(app, host="0.0.0.0", port=50001)