DABEST and permuco

This week’s post is about two very useful libraries for statistical evaluations of behavioral, neuronal, and other types of data, especially when the data deviate from assumptions such as having a Gaussian (aka normal) distribution. (This is often the case with behavioral data.) We hope that you find it useful and the overall awareness of these tools increases.

DABEST is a cross-platform library (Python, R, Matlab) for using estimation stats, developed by Adam Claridge-Chang and colleagues. The approach is a combination of graphics-driven exploratory data analysis, bootstrapped confidence intervals, and permutation (aka randomization) tests. There is extensive documentation of why one should use estimation stats and how to use the methods. Methods exist for the Unpaired Student’s t-test (Two-group estimation plot), the Paired Student’s t-test (Paired estimation plot), One-way ANOVA + multiple comparisons (Multi two-group estimation plot), Repeated measures ANOVA (Multi paired estimation plot), and Ordered groups ANOVA (Shared-control estimation plot). In our experience, DABEST works well for typical datasets from behavioral experiments and little time was needed to incorporate the approach into our data flow. We were literally using the library effectively within a morning given the quality of the documentation.

Another useful library for data analysis is permuco. It was developed by Jaromil Frossard and Olivier Renaud, and is available on CRAN and github. This is a library for R. (It can be used seamlessly in Python notebooks and scripts using the very handy rpy2 library.) permuco allows for permutation testing in regression, ANOVA with interactions among predictors, and comparisons over signals such EEG (or LFP) over time and experimental conditions. A detailed reference manual and vignette are available on CRAN. Similar to DABEST, we have been able to use permuco effectively in a very short period of time, especially because it uses syntax that is common to stock R functions such as aov and lm.

This research tool was created by your colleagues. Please acknowledge the Principal Investigator, cite the article in which the tool was described, and include an RRID in the Materials and Methods of your future publications.  DABEST RRID:SCR_022340; Permuco RRID:SCR_022341