methyLImp2: faster missing value estimation for DNA methylation data

• DOI: 10.1093/bioinformatics/btae001
• Publication Date: 2024-01-12T06:04:48Z
• Authors: Anna Plaksienko; Pietro Di Lena; Christine Nardini; Claudia Angelini
• Keywords: Bioconductor; Imputation (statistics)

Abstract Motivation methyLImp, a method we recently introduced for the missing value estimation of DNA methylation data, has demonstrated competitive performance in data imputation compared to existing, general-purpose, approaches. However, running time was considerably long and unfeasible case large datasets with numerous values. Results methyLImp2 made possible computations that were previously unfeasible. We achieved this by introducing two important modifications have significantly reduced original without sacrificing prediction performance. First, implemented chromosome-wise parallel version methyLImp. This parallelization runtime several 10-fold our experiments. Then, handle datasets, also mini-batch approach uses only subset samples imputation. Thus, it further reduces from days hours or even minutes datasets. Availability implementation The R package is under review Bioconductor. It currently freely available on Github https://github.com/annaplaksienko/methyLImp2.