MScThesis

This project contains the R and python code for my M.Sc. Thesis with the topic "Estimation of higher-order species interactions from ecological time series". It presents a collection of scripts and datasets that are used in the thesis. The repository is structured into explore, Python, and R directories, each containing subdirectories with scripts and data related to the specific field.

Repository Structure

explore Directory

Contains all R code for the preparation of the datasets, including loading the data, cleaning, converting to phyloseq objects, aggregation, filtering, and saving pre-processed time series as CSV files.

• 01a-timeseries-BioTIME: Includes all BioTIME studies.
• 01b-timeseries-CLVpaper: Contains the C. Diff. Bucci dataset.
• 01c-timeseries-miaTIME: Comprises the Silverman Artificial Gut dataset.
• 01d-timeseries-HumanGutData-Karwowska-paper: Encompasses all microbial Human Gut datasets (for the subjects "donorA", "donorB", "male", "female").
• 01e-timeseries-miaSim_files: Describes the simulation of the miaSimS4 time series, including adding noise.
• 01f-timeseries-NODEBNGMpaper: Contains the 3DLV tri-trophic Lotka-Volterra time series and the Ushio fish population dataset.
• 01g-timeseries-simulation-VdP: Describes the simulation of the Van der Pol equation plus adding noise.
• 02-filter_and_group_ts: Structures the process of filtering and grouping the time series and saving them as CSV files.

Python Directory

Contains scripts for alr transformation, the application of compositional Lotka Volterra and DeepMoD, and analysis plus creating plots.

• ALR_transformation: Code for ALR transformation of datasets, along with a folder for all ALR-transformed time series in CSV format.
• CompLotkaVolterra: Provides scripts for both generalized and compositional Lotka-Volterra applications, plus code from the CLV repository.
• DeepMoD: Contains code to run the DeePyMoD framework on time series data.
• Analysis_and_Plots: Includes code for post-modeling analyses and the construction of plots for the thesis.

R Directory

Includes scripts for running the NODEBNGM model and some help functions.

• NODEBNGM: Contains the data and script from https://github.com/WillemBonnaffe/NODEBNGM/tree/main plus two additional scripts: m_main_general.r and run_NODEBNGM.R to run the NODEBNGM algorithm on our time series datasets.
• Functions: Houses helper functions "merge-methods-modified" and "fct_save_time_series".

Getting Started

To use the code and datasets provided in this repository, clone the repo to your local machine using the following command:

git clone https://github.com/mariaproebstl/MScThesis.git

Additional Material: Output Files

Additional output material from the application of generalized and compositional Lotka Volterra, as well as DeepMoD and NODEBGNM to the syntheic and real-world datasets can be found on