Table 1.

Select parameters of Orthrus’ normalization and scoring functions with their associated algorithm (when applicable) and a description of their typical use case.

Parameter	Algorithm	Description	Typical use case
scaling_factor	scaling factor for LFC computation	Scales raw read counts to a default value of 1e6 that forces each screen to the chosen read depth, ensuring comparability across technical replicates. The specific choice of scaling_factor is largely irrelevant	All screens
pseudocount	pseudocount for LFC computation	Adds a pseudocount to each raw read count, by default 1, as required to take log2-normalized read counts. Smaller pseudocounts, e.g. between 1 and 5, are advised to avoid de-prioritizing moderate effects	All screens
test	“moderated-t” - moderated t-testing	Computes p-values via Empirical Bayes estimate across all residuals fit with separate linear models for each gene pair. Calls limma’s eBayes function on its lmFit function applied to residuals with default parameters for both24	Most screens
test	“rank-sum” - Wilcoxon rank-sum testing	Computes p-values via Wilcoxon rank-sum testing between effect and control LFCs	Combinatorial screens with unpaired controls
loess	“TRUE” - Loess normalization with MA transformation	Normalizes by fitting a loess curve with degree 2 and a span of 0.4 to MA-transformed residuals. The MA transformation was originally developed for the analysis of microarray data25. Here, loess fits a trend for the measured residual value ([double mutant - null model] or [condition - control]) vs. the sum of the two values used in computing this residual (eg. [double mutant + null model] or [condition + control]).	Most screens
fdr_method	“BY” - Benjamini-Yekutieli FDR correction	Adjusts p-values with Benjamini-Yekutieli FDR correction	Most screens
fdr_method	“BH” - Benjamini-Hochberg FDR correction	Adjusts p-values with Benjamini-Hochberg FDR correction	Low-signal screens
fdr_method	“bonferroni” - bonferroni FDR correction	Adjusts p-values with Bonferroni multiple hypothesis correction	High-signal screens
filter_genes	N/A	Genes to filter out from scoring process	Remove technical controls or flagged genes
ignore_orientation	N/A	If TRUE, groups guides from both orientations for each gene pair together to reduce the amount of hypothesis testing by half	Cas12a-Cas12a or low-signal screens