Fig 3.

Schematic of pipeline for detecting and correcting chromosomal location bias using ChromoCorrect. Gray boxes indicate user input or output, and white indicates automated steps. Detecting requires the log₂ fold changes of each locus to plot a scatterplot of fold change by chromosomal position, which graphs trends in the fold changes for the user to visualize the pipeline’s assessment. The app and the R console display a message recommending normalization if chromosomal location bias is detected by a fitted linear model. Correcting requires read counts per locus in a txt, csv, or tsv file format, which are then normalized using a sliding window of medians with a default size of 500. An offset matrix is generated from the normalized counts to input along with the raw read counts into edgeR. A linear model is fitted again during correction to determine whether the default sliding window is small enough to capture the trend and repeats the normalization procedure with a smaller window otherwise. The corrected analysis is returned after the normalization is complete.