A database to explore biological relationships between Arabidopsis thaliana genetic characteristics.

Learn More
Feature Name
No records to display

Made with Love

How to use ?

A database to explore biological relationships between A. thaliana genetic characteristics. Search using any feature given in the list of features (link below).


Search by Gene Ontology (GO) terms (e.g GO:0006950) and other A. thaliana feature names above. GO terms (prefix is 'go_') and Differential gene expression (DGE) features (prefix is 'dge_') need to be searched without their prefix. Click here to find out the list of A. thaliana features. Examples of search terms, one per feature category, is given in the table  here. For all features, please click on the link given for the full list. When a feature is searched in the database, the results  show trained machine learning model scores, and network and feature information.

The Out-of-bag (OOB) score (model score) is rounded to 2 decimal places, categorical variables are scored using OOB F1 while  continuous variables are scored using OOB R2.


The network shows the local neighbours of the node (feature). Nodes are coloured according  to their feature categories, edge weight is proportional to the strength of correlation between the nodes, and edge colour is determined  by direction of correlation (red for positive correlation and blue for negative correlation) to the search term  (feature that you have searched for).

The feature information shows feature category, name, description and feature importance information, which is given by the  randon forest model, and shows how strong a feature  is in predicting the search term. It also shows the feature rank score (FRS), which is calculated by converted feature importance values into ranks, and calculating Spearman's correlation from these ranks. Hence, the interpretation of FRS values is  the same as interpreting Spearman's. Its values range from -1 (perfectly negative correlation) to 1 (perfectly positive correlation), with 0  meaning no correation. Feature importance values and FRS scores have been rounded to two decimal places.