01dadada / rt-transformer

RT-Tranformer

The RT-Tranformer combine the fingerprint and the molecular graph data and predict retention time as the output. It's architecture is showed as following:

Motivation

Motivation:Liquid chromatography retention times prediction can assist in metabolite identification, which is a critical task and challenge in non-targeted metabolomics. However, different chromatographic conditions may result in different retention times for the same metabolite. Current retention time prediction methods lack sufficient scalability to transfer from one specific chromatographic method to another

Requirements

• Python 3.9
• torch
• rdkit-pypi
• torch-scatter
• torch-sparse
• torch-cluster
• torch_geometric
• scikit-learn
• tqdm
• jupyter
• notebook
• pandas
• networkx
• gradio

Datasets

The SMRT dataset is collect from this paper Datasets for transfer learning is download from PredRet

Usage

Validation and Test

Run test.py by python ./test.py

Predict Rentention Time

You can follow the jupyter notebook to predict rentention time in your own data.

We provide easily accessible web pages and host them on the huggingface.

Transfer Learn to Your Own Dataset

• Prepare your dataset as a csv file which has "InChI" and "RT" columns.
• Rename it as "data.csv" at the root directory.
• download the pre-trained model from huggingface.
• Run transfer.py

You can also follow this jupyter notebook to fine-tuning the model.

Retrain the Model

• Prepare your dataset as a csv file which has "InChI" and "RT" columns.
• Rename it as "data.csv" at the root directory.
• Run train.py

Pretrained Model Files

best_state_download_dict.pth The Best model of RT-Transformer train from retained data. best_state_dict.pth The Best model of RT-Transformer train from full data.

Cite

If you make use of the code/experiment in your work, please cite our paper (Bibtex below).

@article{xue2023rt, title={RT-Tranformer: Retention Time Prediction for Metabolite Annotation to Assist in Metabolite Identification}, author={Jun Xue and Bingyi Wang and Hongchao Ji and Weihua Li }, year={2023} }