Fan et al. proposed Deep-Learning-Based Multivariate Curve Resolution 2 (DeepResolution2) for automatic resolution of GC-MS data. It utilizes deep neural networks to divide the profile into segments, estimate the number of components in each segment, predict the elution region of each component in overlapped peaks, and resolve the overlapped peaks. The purpose of this repository is to show an example of it modified to work on a set of geologic samples.
python and TensorFlow:
Python 3.6.5 and TensorFlow (version 2.0.0-GPU)
The main packages can be seen in requirements.txt
An example has been provided in Usage.ipynb script for the convenience of users.
Since the model exceeded the limit, Fan et al have uploaded all the models and some GC-MS example data to Zenodo.
https://doi.org/10.5281/zenodo.5498987
Miranda Baker: micabake@ucsc.edu (for questions regarding DeepResolution2 refer to (https://github.com/XiaqiongFan/DeepResolution2).