Effect of imputation on gene network reconstruction from single-cell RNA-seq data

Abstract

Despite the advances in single-cell transcriptomics the reconstruction of gene regulatory networks remains challenging. Both the large amount of zero counts in experimental data and the lack of a consensus preprocessing pipeline for single-cell RNA-seq data make it hard to infer networks from transcriptome data. Data imputation can be applied in order to enhance gene-gene correlations and facilitate downstream data analysis. However, it is unclear what consequences imputation methods have on the reconstruction of gene regulatory networks. To study this question, we evaluate the effect of imputation methods on the performance and structure of the reconstructed networks in different experimental single-cell RNA-seq data sets. We use state-of-the-art algorithms for both imputation and network reconstruction and evaluate the difference in results before and after imputation. We observe an inflation of gene-gene correlations that affects the predicted network structures and may decrease the performance of network reconstruction in general. Yet, within the modest limits of achievable results, we also make a recommendation as to an advisable combination of algorithms, while warning against the indiscriminate use of imputation before network reconstruction in general.