Cyclical Regression Covariates remove the major confounding effect of cyclical developmental gene expression with strain-specific drug response in the malaria parasitePlasmodium falciparum

Abstract

40Background: The cyclical nature of parasite gene expression in the intraerythrocytic development cycle 41 (IDC) in human blood confounds the accurate detection of specific transcriptional differences due to 42 drug resistance in Plasmodium falciparum. Here, we propose the use of cyclical regression covariates to 43 eliminate the major confounding of developmentally driven transcriptional changes with changes due to 44 drug response. We show that elimination of this confounding can reduce both Type I and Type II errors, 45and demonstrate the effect of approach on real data. 46Results: We apply this method to two publicly available datasets, and demonstrate its ability to reduce 47 the potential confounding of differences in expression due the species-specific intraerythrocytic 48 development cycle from strain-specific differences in drug response. We show that the application of 49 cyclical regression covariates has minimal impact on the pool of transcripts identified as significantly 50 different in a dataset generated from single timepoint clinical blood samples with low variance for 51 developmental stage and a profound impact on another clinical data set with more variance among the 52 samples for developmental stage. 53Conclusions: Cyclical regression covariates have immediate application to studies where in-vitro 54 synchronization of all samples to the same developmental timepoint is not feasible, primarily parasite 55 transcriptome sequencing direct from clinical blood samples, a widely used approach to frontline 56 detection of emerging drug resistance. 57 58