Table 8. Troubleshooting Table.

Common screen quality issues reflected by Orthrus’ output as well as common issues encountered by users.

Procedure - Step	Problem	Possible reasons	Solutions
2–7, 3–4	Many guides with too few reads in early-timepoint screens are filtered out	Various experimental issues early on in the screen, such as with T0 samples	Orthrus’ scoring accounts for this, but users may consider applying the “ignore_orientation” flag during scoring
2–9, 3–4	Technical replicates correlate poorly with each other	Mislabelling in the sample file or any number of experimental issues	Fix technical replicate labels in the sample file, consider filtering early-timepoint guides more stringently, or redo the problematic screen
2–9, 3–4	Guides appear to drop out stochastically between replicates in QC scatterplots	Dosage for a drug screen was too high	Remove guides that completely drop out in any late-timepoint replicate, or redo the problematic screen
2–5, 3–4	The function plot_reads_qc returns the error “Error in check_screen_para ms(df, screens) : replicate [REPLICATE] not in df, remove screen [REPLICATE] with remove_screens”	Replicate name set in add_screen function does not exist in input dataframe	Remove screen containing the offending replicate from the screen list with the function remove_screen and readd with correct replicate names
2–11, 2–15, 2–18, 3–4	Many rows in scored data contain NA values	Guide filtering based on early-timepoint readcounts removed too many guides to score gene pairs with NA values (gene pairs with fewer guides than the min_guides parameter are not scored)	Consider relaxing guide filtering threshold, lowering the min_guides parameter, or investigating issues in early-timepoint screens
2–11, 2–15, 2–18, 3–4	Scoring takes much longer than expected given the number of screens	Control gene pairs with many guides are not filtered out	Add controls such as “NT” guides to the filter_genes scoring parameter in a vector