I'm using Anaquin to measure detection of Sequin fold change by a sequencer. The fold change is between Sequin mixes A and B. When I try to make a ROC plot using the following parameters: - Expected log fold change - 1 minus P-value of measurement - TP/FP label of measurement plotROC collapses all the curves, each one representing a different level of fold change, into a single curve. Adding to the problem the curve is also a "perfect" right-angle. I've tried using multiple definitions of a true positive including P < 0.05, P < 0.001 and Q < 0.05, but I still have the issue.
Here's the code I've been using:
# Title title <- 'ROC gene plot - Illumina' # Sequin names seqs <- row.names(exp_gene_sequins) # Expected log fold ratio <- exp_gene_sequins$Int_Log2_Fold_change # How the ROC points are ranked (scoring function) score <- 1 - illumina_gene_sequins$pvalue # Classified labels (TP/FP) label <- illumina_gene_sequins$Label # Make the plot plotROC(seqs, score, ratio, label, title=title, refGroup=0, legTitle = "Fold change")
Summary plotROC function bunches all fold changes together into a single line
Many thanks, Will