Table 3.

Publicly available datasets used for computationally identifying BACE1 substrate lists.

Reference	Study Type	Experimental Conditions	How Data Was Used	Statistical Analysis by Authors
[11]	SILAC proteomics	Brain membrane fraction from WT and BACE1 KO mice	Identify BACE1-regulated proteins	Log2 ratios of technical replicates were averaged and average protein log2 fold changes were calculated between BACE1 KO and WT samples. A two-sided Student’s t test evaluated the significance of proteins. Permutation-based FDR estimation was used.
[30]	SPECS proteomics	Primary neurons from BACE1 inhibitor treated, WT, and BACE1 KO mice	Identify BACE1-regulated proteins	A variance score (VS = absolute value of (standard error of the mean/(1 − mean))) was calculated for all proteins. Proteins with a vs. of ≤0.35 were considered as proteins with a consistent change upon BACE1 inhibition.
[31]	Loss/Gain-of-Function assays	MIN6 with knockdown of BACE1 and BACE2	Identify BACE1-regulated proteins	Protein significance analysis was performed using SRMstats where a constant normalisation was performed for all runs to equalise the median peak intensities of the heavy transitions from all the peptides between runs.
[32]	Quantitative proteomics	CSF from WT and BACE1 −/− mice	Identify BACE1-regulated proteins	Using the mean, the average LFQ intensity within each biological group was calculated. A two-sided t test was used, and the p value was corrected using false discovery rate (FDR)-based multiple hypothesis testing.
[33]	SILAC proteomics	HeLa/HEK with BACE1 overexpression	Identify BACE1-regulated proteins	Proteins containing peptides with at least 65% of the total signal derived from the BACE1 condition were considered as putative substrates. This threshold value is equivalent to a 1.857-fold increase in peptide abundance.
[34]	BACE1 substrate prediction	Bioinformatic analysis	Identify potential BACE1 substrates
[35]	Single nucleus RNA sequencing	Single nucleus prefrontal cortical samples of AD patients and normal control (NC) subjects	Used to identify genes differentially expressed in Alzheimer’s disease	Data were background corrected and quantile normalised. Differential expression was performed via limma using a block design to leverage technical replicates. Genes with a false discovery rate (FDR)-adjusted p < 0.05 were considered differentially expressed. An algorithm based on the hypergeometric distribution is used to calculate enrichment p values.
[36]	RNA sequencing	Endothelial and brain cells from mouse models of stroke, multiple sclerosis (EAE), traumatic brain injury (TBI), and epilepsy	Identify genes differentially expressed in endothelial cells, dependent on disease	Stratified samples according to classification of AD and NC samples and compared the transcriptome profiles of individual cell types between AD and NC samples by the Wilcoxon rank-sum test using the FindMarkers function with the parameters logfc.threshold = 0 and test.use = wilcox. The level of statistical significance for cell-type-specific transcriptomic changes was set at an adjusted p < 0.1 and a log2 fold change ≥0.1 or ≤−0.1.