Statistical Report

To compare or analyse a model after curation, some kind of statistis can be very handy. One idea is to analyse the model with memote, however, the report consists - while being very detailed - purely of numerical values. Additionally, if running it multiple times is required, it can be quite time consuming.

The toolbox refineGEMs provides a quick and graphic alternative in form of the ModelInfoReport class. It can produce a report on the main (statistic) properties of a model, including:

• The basic counts of reactions, metabolites and genes

• Counts of the types of metabolites in the model

• Number of reactions with and without GPRs

• Number and types of unbalanced reactions

Furthermore, these values can be visualised as bar and doughnut chart.

Note

We are currently working on an extension of this class that directly produces a report that compares multiple models instead of just analysing one.

How to create the ModelInfoReport

Via command line

The basic command is:

refinegems analyse stats MODELPATH

Additionally, the path to the output directory can be added using the flag --dir/-d and the colours of the plot can be changed by passing a valid matplotlib colour palette abbreviation to --colors/-c.

Inside Python

Assuming a model variable model that contains a cobra.Model entity exists, the report can be generated as follows:

report = ModelInfoReport(model)

fig = report.visualise() # produces the graphic

# dir : Path to output directory
report.save(dir) # save the report (graphic + table)

Examplary ModelInfoReport visualisation

An exemplary visualisation of a report on a Klebsiella pneumoniae model for a strain from a private collection is shown below.

The visualisation contains three subfigures, one for the overview (upper left corner), one for further statistics about the metabolites (upper right) and one for more information about the reactions in the model (bottom). The colours are from the default colour palette YlGn.