Code to wrangle Raman spectra, visualize Raman spectra, identify components in a mixture Raman spectra, and ultimately identify decomposition or presence of materials using Raman spectroscopy.
Team members (alphabetical by first name): Brandon Kern, Elizabeth Rasmussen, Parker Steichen
This project identifies, and calculates decomposition in a Raman spectra to output rate data. Advantages to using this method are:
- FULLY Open source, no part of the project is dependent on a paid service
- AUTOMATED process, analysis is automated leading to fast results
- VERIFIABLE, user is made aware of how confident they can be in the results via a statistical software stack
- More detailed background and motivation for the project can be found in the Project Background and Motivation Wiki Page
- Storing data library that is beyond the decomposition products of formic acid (hydrogen, water, carbon dioxide, carbon monoxide) are not included as other components are beyond the scope of the project at this time.
- Assumed that the user is trying to analyze the decomposition products of formic acid, or a mixture that consists only of: Formic Acid, Hydrogen, Water, Carbon dioxide, Carbon monoxide.
- This project will not be predictive - that is, it will require the user to specifically input what potential compounds will be present in the spectra to be analyzed. This list does not have to be exhaustive; however, the more inclusive the list, the better the fitting and predicting results will be.
The project can be thought of as broken down into 3 steps:
- Data Wrangling
- Peak fitting and identification
- Statistical analysis for peak fits
These sections have their own wiki documents and filled juypter notebooks with more detail included throught, see those for more detail on the individual steps.
A user will be able to follow the steps to apply Raman-Noodles to YOUR Formic Acid data set. An example of using the software can be seen in the Example Use Case Wiki Page
In order to have manageable code we are using Travis' Open Source continuious itegration testing. One thing to note is that on March 1, 2018 Travis-CI switched their model for open source software, the press release about this can be read here
So we do have our team repo viewable on travis-ci.com BUT it will ultimately re-direct you to the old platform for open source software on travis-ci.org.
There is a way to merge the travis-ci.org (open source repos only) to travis-ci.com (now private repos and (closed beta) open source repos) as can be seen by following this link but at this time our team has decided to not join the closed beta as the current (old) method of the dashboard being located on travis-ci.org works just fine.
Next steps would include automatic baseline subtraction of spectra to decrease pre-processing time, molar formation calculations to predict reaction pathways, and increased robustness of machine learning for component selection including unsupervised methods.
In conclusion our team successfully created a platform code base for researchers to visualize and analyze their Raman spectra in a fast, automated manor - reducing post-processing time by days and enabling future work to continue on a solid base of open source tools.
-
This software has passed tests to sucessfully identify and analyze the identification of components in mixture Raman data.
-
This work sets up a free and user friendly platform for researchers to be able to analyze their own Raman Spectra.
- Project Sponsor: Igor Novosselov
- Additional assistance: Dave Beck, and Kelly Thornton
- Data sets from publicly available from Mendeley Data, “Raman Spectra of Formic Acid Gasification Products in Subcritical and Supercritical Water”
- Only open source packages were used in this work.