Skip to content

Latest commit

 

History

History
87 lines (71 loc) · 3.8 KB

R2APP.md

File metadata and controls

87 lines (71 loc) · 3.8 KB
Raw

R Package converting to BioMANIA APP README

This README framework provides a step-by-step guide for integrating R packages into the BioMANIA app. The process is streamlined for R, leveraging its capabilities for direct package usage and documentation access. Python scripts are still used for data preparation and model training.

Step 1: Prerequisites

Before starting, ensure that R and rpy2 are installed. rpy2 allows for running R within Python, bridging the two environments.

  • Install R: Download and install the latest version of R from CRAN.
  • Install rpy2: In your Python environment, run
pip install rpy2

Step 2: Generating API Initialization File

Use a Python script to generate the ./data/R/{Your_R_Library_Name}/API_init.json file. This file contains initial information about the APIs in the R package.

  • Run the following command:
python dataloader/get_API_init_from_sourcecode_R.py --LIB [Your_R_Library_Name]
cp -r ./data/R/{Your_R_Library_Name}/API_init.json /data/R/{Your_R_Library_Name}/API_composite.json

Step 3: Preparing Retriever Data

Prepare the data required for the retriever model using another Python script. This process generates the instruction under ./data/R/{your_r_lib_name}/API_inquiry.json file and ./data/R/{your_r_lib_name}/retriever_train_data/ folder.

  • Execute the command:
python dataloader/prepare_retriever_data.py --LIB [Your_R_Library_Name]

Step 4: Training Models

Train the chat classification and retriever models. These models are crucial for the app's functionality.

  1. Chitchat Classification Model:
  • Train the model using the following command:
python models/chitchat_classification.py --LIB ${LIB}
  1. Retriever Model:
  • Train the model with this command (adjust the command with necessary parameters):
export LIB=scanpy
CUDA_VISIBLE_DEVICES=0
mkdir ./hugging_models/retriever_model_finetuned/${LIB}
python models/train_retriever.py \
--data_path ./data/standard_process/${LIB}/retriever_train_data/ \
--model_name bert-base-uncased \
--output_path ./hugging_models/retriever_model_finetuned/${LIB} \
--num_epochs 25 \
--train_batch_size 32 \
--learning_rate 1e-5 \
--warmup_steps 500 \
--max_seq_length 256 \
--optimize_top_k 3 \
--plot_dir ./plot/${LIB}/retriever/

Step 5: Use model

Start back-end UI service with:

export LIB=ggplot2
CUDA_VISIBLE_DEVICES=0 \
python deploy/inference_dialog_server_R.py \
    --retrieval_model_path ./hugging_models/retriever_model_finetuned/${LIB}/assigned \
    --top_k 3

Install and start the front-end service in a new terminal with:

cd src/chatbot_ui_biomania
npm i # install
export BACKEND_URL="https://[ngrok_id].ngrok-free.app" # "http://localhost:5000";
npm run dev # run

Your chatbot server is now operational at http://localhost:3000/en, primed to process user queries.

Key Differences Between R and Python Integration

  • Library Loading: In R, use library(LIB) to load packages directly. There's no need to modify Lib_cheatsheet.py as in Python.
  • Documentation Access: R documentation can be accessed through help(), ??, or the .__doc__ attribute after converting R functions to Python via rpy2.
  • Arguments Information: R documentation didn't always provide type information for parameters.
  • Simplified Process: The process for R integration is more straightforward, focusing primarily on data preparation and model training, without the need to adjust library settings extensively.

Final Notes

This framework outlines the key steps and differences for integrating an R package into BioMANIA. Adjust the Python commands and paths according to your package's specifics. If you have any questions or need assistance with specific steps, feel free to reach out!