R.ROSETTA: an interpretable machine learning framework

Abstract

Background Machine learning involves strategies and algorithms that may assist bioinformatics analyses in terms of data mining and knowledge discovery. In several applications, viz. in Life Sciences, it is often more important to understand how a prediction was obtained rather than knowing what prediction was made. To this end so-called interpretable machine learning has been recently advocated. In this study, we implemented an interpretable machine learning package based on the rough set theory. An important aim of our work was provision of statistical properties of the models and their components. Results We present the R.ROSETTA package, which is an R wrapper of ROSETTA framework. The original ROSETTA functions have been improved and adapted to the R programming environment. The package allows for building and analyzing non-linear interpretable machine learning models. R.ROSETTA gathers combinatorial statistics via rule-based modelling for accessible and transparent results, well-suited for adoption within the greater scientific community. The package also provides statistics and visualization tools that facilitate minimization of analysis bias and noise. The R.ROSETTA package is freely available at https://github.com/komorowskilab/R.ROSETTA. To illustrate the usage of the package, we applied it to a transcriptome dataset from an autism case–control study. Our tool provided hypotheses for potential co-predictive mechanisms among features that discerned phenotype classes. These co-predictors represented neurodevelopmental and autism-related genes. Conclusions R.ROSETTA provides new insights for interpretable machine learning analyses and knowledge-based systems. We demonstrated that our package facilitated detection of dependencies for autism-related genes. Although the sample application of R.ROSETTA illustrates transcriptome data analysis, the package can be used to analyze any data organized in decision tables.

Transcriptomic analysis reveals proinflammatory signatures associated with acute myeloid leukemia progression

“…After selecting the most important features from high-dimensional data, interpretable machine learning was performed by using the R.ROSETTA framework. 26 Subsequently, rule-based models were constructed, and copredictive features for each data set were estimated.…”

“…To create a ranking of the most informative features (ie, genes) that distinguish between diagnosis and relapse, the Monte Carlo Feature Selection (MCFS 25 ; rmcsf v.1.2.6) algorithm was applied. Subsequently, interpretable machine learning models were built by using the R.ROSETTA R-package 26 version 2.2.9. That algorithm is based on the rough set theory.…”

Interpretable machine learning identifies paediatric Systemic Lupus Erythematosus subtypes based on gene expression data

“…The initial rule-based model was built with R.ROSETTA 13 using data from 629 unique patient clinical visits (observations) and the discretised gene expression value for each DA1 and DA3 patient visit (features: 33,006 probes for 629 observations; Figure 1). This initial model had an overall prediction accuracy of 71% using 10-fold cross validation (Supplementary Fig S1 online).…”

“…First, expression values were subject to data discretisation, since R. ROSETTA 13 generates rules for that data form. For each gene, the control data expression mean (μ) and standard deviation (σ) were calculated, and then all DA data for that gene projected onto this threshold frame and discretised (Low ≤ μ - 2σ < Medium > High ≥ μ - 2σ; Numeric values 1, 2, 3).…”

“…The significant features from MCFS output were incrementally added to build several rule-based models. The selected features that were used to build the model with the best overall accuracy where chosen for building the final enhanced model using R.ROSETTA 13 and then visualized using VisuNet 41 .…”