Skip to content

Lapis (v2023.9.12)

Latest
Latest
Compare
Choose a tag to compare
@sunroofgod sunroofgod released this 12 Dec 06:00
v2023.9.12
a4ddde6
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Exciting news for dabestr users! We are releasing dabestr version “Lapis” (v2023.9.12), which adds estimation graphics for four new data types: repeated measures, 2 × 2 designs, proportions, and easy meta-analyses.

  1. First up, in dabestr Lapis, a new graphic is suited for repeated-measures effect sizes. You no longer are stuck running repeated-measures ANOVAs, but can instead visualize best-practice estimation plots. Repeated-measures plots enable you to see a series of effect sizes relative to a baseline measurement.
  2. Instead of using a 2-way ANOVA, you can now generate a ∆∆ plot (‘delta-delta plot’) to visualize a 2 × 2 experiment. The ∆∆ shows individual effect sizes for each of the two-groups sub-comparisons, and the overall magnitude of the interaction effect. For example, ∆∆ plots will be useful for genotype × drug analyses or optogenetic (light × opsin) experiments. A major benefit of the ∆∆ plot is how it places focus on the overall effect size.
  3. Instead of Fisher’s exact test for the comparison of proportions, you can now use the new DABEST proportion plots. It plots both the variance of the observed binary counts and the effect-size curve of the difference of proportions. This function includes repeated measures of proportions.
  4. Last but not least, we introduce the mini-meta plot, a quick way to do meta-analytic averaging from similar experiments. This provides an easy way to summarize results from multiple replicates. This makes meta-analysis of primary data available as a convenient tool.

We have also added several features not yet present in the Python version of the package:

  1. We have adopted asymmetrical swarmplots, thereby better showing the distribution underlying the raw data.
  2. Users may add a baseline error curve to shared control contrast plots. This is the distribution of the bootstrapped mean differences from comparing the control against itself, and is a measure of inherent variability within the control group.

Updated documentation for the R package can be found at: https://acclab.github.io/dabestr/

Contributers to this update were: Kah Seng LIAN (@sunroofgod), Zhuoyu WANG (@Lucas1213WZY) and Jun Yang LIAO (@junyangliao)

Contributors

sunroofgod, junyangliao, and Lucas1213WZY
Assets 2