This repository contains code and data (or link to needed data) needed to replicate the analysis carried out in the pre-print by Li G, et al, 2019 (https://arxiv.org/abs/1912.08141).
The different parts of the analysis is broken down into separate folders:
-
MonteCarloSimulationcontains the script for Monte Carlo simulations in Figure 2. -
EvaluateClassicalModelscontains the script to evaluate the performance of six different classical regression models on a given dataset. This script was used in the enzyme Topt prediction section (Figure 3). -
EnzymeToptcontains the two fasta files for enzyme Topt: ones with raw and cleaned Topt records. It also contains the script for feature extraction (Figure 3). -
DeepNNcontains the architecture of deep neural networks used in the prediction of enzyme Topt (Figure 3). -
Transcriptomicscontains scripts for the transcriptomics data analysis (Figure 4). -
Genomicscontains the scripts, datasets and results for the evaluation of a random forest model on the prediction 34 quantitative phenotypes from pan genome (Figure 5). The pan genome data was obtained from https://zenodo.org/record/3407352#.Xe-gX5P0nUI. Quantitative phenotypes were obtained from Peter, J. et al. Genome evolution across 1,011 Saccharomyces cerevisiae isolates. Nature 556, 339–344 (2018).
numpy 1.15.0
pandas 0.23.4
sklearn 0.20.3
seaborn 0.9.0
keras 2.2.4
tensorflow 1.10.0
hyperopt 0.1.2
The repository was tested with Python 3.6.7.
Scripts of MonteCarloSimulation and Transcriptomics can be performed with PC with linux or Mac OS. Other scripts need to be done with a computer cluster. All deep learning related analysis (UniRep representation and DeepNN) need GPU platform.