AutoRes

KO | EN

gitlite — search

#ai #llm #golang #deep-learning #linux #docker #java #windows #awesome #vue #awesome-list #pytorch #openai #mcp

(★ 9)

Deep Learning-based Method for Automatic Resolution of GC-MS Data from Complex Samples

AutoRes 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## AutoRes

This is the code repo for the paper *Deep Learning-based Method for Automatic Resolution of GC-MS Data from Complex Samples*. We proposed an Automatic Resolution method (AutoRes) for overlapped peaks in complex GC-MS data based on the pseudo-Siamese convolutional neural network (pSCNN) architecture. It consists of two pSCNN models (pSCNN1 and pSCNN2) with the same architecture but different inputs. Two pSCNN models were trained with 400,000 augmented spectral pairs, respectively. They can predict the selective region (pSCNN1) and elution region (pSCNN2) of compounds in an untargeted manner. The predicted regions are used as inputs to the full rank resolution (FRR) method, which can be easily achieved for the overlapping peaks.

### Package required:
We recommend to use [conda](https://conda.io/docs/user-guide/install/download.html) and [pip](https://pypi.org/project/pip/).
- [python3](https://www.python.org/)
- [tensorflow](https://www.tensorflow.org)

By using the [`environment.yml`](https://github.com/dyjfan/AutoRes/blob/main/environment.yml), [`requirements.txt`](https://github.com/dyjfan/AutoRes/blob/main/requirements.txt) file, it will install all the required packages.
    
    git clone https://github.com/dyjfan/AutoRes.git
    cd AutoRes
    conda env create -f environment.yml
    conda activate autores

## Data augmentation

The mass spectral pairs of the training pSCNN1 and pSCNN2 models are obtained using the [data_augmentation_1](https://github.com/dyjfan/AutoRes/blob/main/pSCNN/da.py#L23) and [data_augmentation_2](https://github.com/dyjfan/AutoRes/blob/main/pSCNN/da.py#L47) functions.

aug_eval1 = data_augmentation_1(spectra, n, maxn, noise_level=0.001)
    aug_eval2 = data_augmentation_2(spectra, c, n, m, maxn, noise_level=0.001)

*Optionnal args*
- spectra : Mass spectral library 
- c : Similar sublibrary
- n ：Number of amplified mass spectral pairs
- m : Number of amplified mass spectral pairs with high similarity.
- maxn ：Number of components

## Model training
Train the model based on your own training dataset with [build_pSCNN](https://github.com/dyjfan/AutoRes/blob/main/pSCNN/snn.py#L69) function.

model = build_pSCNN(para)

*Optionnal args*
- para : Hyperparameters for model training

## Automatic Resolution

Automatic Resolution of GC-MS data files by using the [AutoRes](https://github.com/dyjfan/AutoRes/blob/main/AutoRes/AutoRes.py#L633) function.

AutoRes(ncr, model1, model2, filename)
    
*Optionnal args*
- ncr : GC-MS data
- model1 : pSCNN1 model
- model2 : pSCNN2 model
- filename : GC-MS data filename

## Clone the repository and run it directly
[git clone](https://github.com/dyjfan/AutoRes)

An example has been provided in [test.ipynb](https://github.com/dyjfan/AutoRes/blob/main/test.ipynb) script for the convenience of users. Users can run it directly by placing the unzipped data file and AutoRes-1.0 file in the same directory after downloading.

It should be noted that the NIST_Spec.db database is not available for download due to copyright reasons. Users who need to use the database should obtain it from NIST.

## Contact
- fanyingjie@csu.edu.cn

AutoRes

AutoRes 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

AutoRes

AutoRes 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README