mSOM (metaboliteSOM)
John Pearce
MSc Applied Bioinformatics 2023
Cranfield University
[email protected]
[email protected]
- Investigate the suitability of conventional clustering methods (K-means) for metabolic time-series data
- Apply and optimise SOMs for the same data set
- Compare K-means with a Self-Organizing Map (SOM)
- Evaluate the appropriateness of conventional measures to assess cluster quality
- Investigate the use of additional biological information (common metabolic pathways) to classify types of clusters.
The workflow is illustrated below. It consists of a Conventional Processing Workflow and a SOM Clustering & Biological Connectivity Workflow. The former provides basic data validation and then performs data exploration (elbow graphs, clustering metrics, K-means and PCA). The latter provides SOM clustering (using minisom package) and Biological Connectivity (using additional data on common metabolic pathways).
A time-series metabolic data set (Arabidopis thaliana) was obtained from Kim, Jae Kwang & Cho, Myoung & Baek, Hyung-Jin & Ryu, Tae & Yu, Chang & Kim, Myong & Fukusaki, Eiichiro & Kobayashi, Akio. (2007). Analysis of metabolite profile data using batch-learning self-organizing maps. Journal of Plant Biology. 50. 517-521. 10.1007/BF03030693.
Any other metabolic time-series data file can be used. The format is a CSV file with a metabolite per row, and each row contains a number of readings at specified points in time.
Additional biological data (common metabolic pathways) was obtained (for Arabidopsis thaliana) from the BioCyC database using the PythonCyC package.
A Data folder contains the required data files. Including the additional biological data if you don't want to install and run Pathway-tools (part of the BioCyC server).
The Conventional Processing Workflow is implemented in DataExploration.ipny and Advanced KmeansAnalysis.ipny
The SOM clustering and Biological Connectivity workflow is implemented in SOMClustering.ipny, CommonPathwayData.ipny and BiologicalConnectivity.ipny
Additional comments are provided in the respective .ipny files.
If you want to run CommonPathway.ipny you will need to get a licence for BioCyC from https://biocyc.org/download-bundle.shtml and install it locally (or on a server). You'll also need to download the the PGDB file for Arabidopsis thaliana (or any other species). Instructions for installing pathway-tools can be found at https://biocyc.org/download.shtml
Alternatively the common pathway data file for Arabidopsis thaliana is included in the Data folder.
SOMs are implemented using the minisom package (https://github.com/JustGlowing/minisom).
A significant number of Python packages are used so it is recommended that a virtual environment like Anaconda is used. A YAML file (environment.yaml) is provided in the repository to help with the construction of an appropriate environment. The appropriate environment can then be created with the command conda env create -f environment.yml