Metabolic Tampering (MeTa) repository

Documentation and models from Facilitating NMR Resonance Assignment with Metabolic Tampering.

Authors: Danica Cui, Evan Anderson, Erik Zabala, George Lisi, J. Patrick Loria

Note Although many of the notebooks are commented and annotated, I am still working on some final annotations. If you have any questions, please don't hesitate to reach out!

In our upcoming publication Facilitating NMR Resonance Assignment with Metabolic Tampering, we describe a new method to aid in assignment of amino acids in NMR experiments. As a part of this publication, we describe implementation of a random forests classifier to classify amino acid identify in 2D ¹⁵N HSQC experiments with varying, small amounts of enriched LB media. This repository serves as documentation for implementation and validation of this model, and also provides templates for implementation of this model by other users.

File Summary

Requirements and environment

We used Python 3.7 for these analyses. requirements.txt lists the Python packages used in this analysis, and environment.yml is the conda environment file which I used for all of the following Jupyter Notebooks.

Data processing, model selection, and validation

meta_documentation.ipynb: Jupyter Notebook describing data preprocessing, model selection, and model validation.

comparing_training_sets.ipynb: Jupyter Notebook describing performance of the model with different train/test splits, including training on one protein (PTP1B) and testing on another (IGPS).

Notebooks for model implentation

retrain_model.ipynb Jupyter notebook serving as a template for training a random forest classifier model on your own data. prepping sample data.ipynb was used to create the place-holder data in this file.

all_unlabeled_model.ipynb Jupyter notebook serving as a template for using the random forest classsifier trained on PTP1B and IGPS on your own unlabeled data.

Data

This folder contains all of the data used in all of the jupyter notebooks in this repo. The datasets used to train the random forests classifier for our publication were: IGPS_8hr_0513.csv PTP1B_8hr_0513.csv

Model

This folder contains the model which we trained meta_documentation.ipynb and which is used in all_unlabeled_model.ipynb.

Output

This folder contains the sample output data files from retrain_model.ipynb and all_unlabeled_model.ipynb, specifically with predicted labels for each peak.