Analysis of time resolved spectra using a deep Spectrosopy Kinetic Analysis Network (deepSKAN) and Global/Target Analysis. If this package contributes to a scientific publication, please acknowledge this by citing https://doi.org/10.1021/acs.jpcb.0c04299
- This package is meant to be installed with pip:
pip install .
The versions specified in requirements.txt are known to work, newer versions will probably be fine but have not been tested
- Python 3.6
- keras (using tensorflow backend)
- networkx
- numpy
- scipy
- matplotlib
In the examples folder, there are three different ways this package can be used:
- Training: Trains a resNet-like CNN on synthetic training data generated on the fly.
- Complete Analysis: Predicts model using neural network, then performs GTA with predicted model.
- Suggestion: gives suggestions about possible pathways. this is implemented as a webinterface.
Basic descriptions of every function are included in their docstring.