Table 2 R Functions and parameters used for NanoString data analysis. See Methods for in-depth description of the procedures.

RCC file import

counts = parse_counts(RCC file), attributes = parse_attributes(RCC file)

nanostringr (0.1.0)

CPM normalization

calcNormFactors(DGEList object, method = “none”)

edgeR (3.40.2)

Filtering of lowly expressed genes

filterByExpr(DGEList-object, group = disease subtype, min.count = 2^(mean(per-probe average negative-control count) + 2 * sd(per-probe average negative-control count)))

edgeR (3.40.2)

CodeSet Content Normalization

in-house script (Supp. Information)

R base (4.2.3)

Design matrix creation

model.matrix(~ 0 + disease subtype + cartridge + conservation)

R stats (4.2.3)

Heteroschedasticity correction (voom)

voom(calcNormFactors(DGEList object, method = “TMM”), design = model.matrix(…), normalize.method = “none”, plot = F)

limma (3.54.0)

Intersample correlation estimation

duplicateCorrelation(voom output, model.matrix(…), block = Patient-ID)$consensus.correlation

limma (3.54.0)

Linear model regression

lmFit(voom output, model.matrix(…))

limma (3.54.0)

Contrast matrix creation

in-house script (Supp. Information)

R base (4.2.3)

Fold-change estimation and statistical inference (moderated t-test, F-test)

contrasts.fit(lmFit output, contrast matrix); eBayes(contrasts.fit(…),trend = F,robust = F)

limma (3.54.0)

Statistical inference (post hoc moderated t-test)

decideTests(eBayes output,method = ‘hierarchical’, adjust.method = ‘BH’, p.value = .05))

limma (3.54.0)

APSPs identification

all_shortest_paths(Reactome graph, from = i, to = j, mode = “all”) for any i, j in gene list

igraph (1.3.5)

Correlation test

cor.test(i, j, alternative = "two.sided", method = “spearman”, continuity = T) for any i, j in gene list

R stats (4.2.3)

Gene sets and pathway resources

Cypher queries (Supp. Information)

docker.io/reactome/graphdb (version 90)