tsea_mabs: Target Set Enrichment Analysis (TSEA) with meanAbs
In signatureSearch: Environment for Gene Expression Searching Combined with Functional Enrichment Analysis

Description Usage Arguments Details Value Column description References See Also Examples

The meanAbs (mabs) method is a simple but effective functional enrichment statistic (Fang et al., 2012). As required for TSEA, it supports query label sets (here for target proteins/genes) with duplications by transforming them to score ranked label lists and then calculating mean absolute scores of labels in label set S.

tsea_mabs(
  drugs,
  type = "GO",
  ont = "MF",
  nPerm = 1000,
  pAdjustMethod = "BH",
  pvalueCutoff = 0.05,
  minGSSize = 5,
  maxGSSize = 500,
  dt_anno = "all",
  readable = FALSE
)

`drugs`	character vector containing drug identifiers used for functional enrichment testing. This can be the top ranking drugs from a GESS result. Internally, drug test sets are translated to the corresponding target protein test sets based on the drug-target annotations provided under the `dt_anno` argument.
`type`	one of 'GO', 'KEGG' or 'Reactome'
`ont`	character(1). If type is 'GO', assign `ont` (ontology) one of 'BP','MF', 'CC' or 'ALL'. If type is 'KEGG' or 'Reactome', `ont` is ignored.
`nPerm`	integer, permutation number used to calculate p-values
`pAdjustMethod`	p-value adjustment method, one of 'holm', 'hochberg', 'hommel', 'bonferroni', 'BH', 'BY', 'fdr'
`pvalueCutoff`	double, p-value cutoff
`minGSSize`	integer, minimum size of each gene set in annotation system
`maxGSSize`	integer, maximum size of each gene set in annotation system
`dt_anno`	drug-target annotation source. Currently, one of 'DrugBank', 'CLUE', 'STITCH' or 'all'. If 'dt_anno' is 'all', the targets from the DrugBank, CLUE and STITCH databases will be combined. Usually, it is recommended to set the 'dt_anno' to 'all' since it provides the most complete drug-target annotations. Choosing a single annotation source results in sparser drug-target annotations (particularly CLUE), and thus less complete enrichment results.
`readable`	TRUE or FALSE, it applies when type is 'KEGG' or 'Reactome' indicating whether to convert gene Entrez ids to gene Symbols in the 'itemID' column in the result table.

The input for the mabs method is L_tar, the same as for mGSEA. In this enrichment statistic, mabs(S), of a label (e.g. gene/protein) set S is calculated as mean absolute scores of the labels in S. In order to adjust for size variations in label set S, 1000 random permutations of L_tar are performed to determine mabs(S,pi). Subsequently, mabs(S) is normalized by subtracting the median of the mabs(S,pi) and then dividing by the standard deviation of mabs(S,pi) yielding the normalized scores Nmabs(S). Finally, the portion of mabs(S,pi) that is greater than mabs(S) is used as nominal p-value (Fang et al., 2012). The resulting nominal p-values are adjusted for multiple hypothesis testing using the Benjamini-Hochberg method.

feaResult object, the result table contains the enriched functional categories (e.g. GO terms or KEGG pathways) ranked by the corresponding enrichment statistic.

The TSEA results (including tsea_mabs) stored in the feaResult object can be returned with the result method in tabular format, here tibble. The columns in this tibble specific to the mabs method are described below.

mabs: given a scored ranked gene list L, mabs(S) represents the mean absolute scores of the genes in set S.
Nmabs: mabs(S) normalized

Additional columns are described under the 'result' slot of the feaResult object.

Fang, Z., Tian, W., & Ji, H. (2012). A network-based gene-weighting approach for pathway analysis. Cell Research, 22(3), 565-580. URL: https://doi.org/10.1038/cr.2011.149

feaResult, fea, tsea_mGSEA

data(drugs10)
## GO annotation system
#res1 <- tsea_mabs(drugs=drugs10, type="GO", ont="MF", nPerm=1000, 
#                  pvalueCutoff=0.05, minGSSize=5)
#result(res1)
## KEGG annotation system
#res2 <- tsea_mabs(drugs=drugs10, type="KEGG", nPerm=1000, 
#                  pvalueCutoff=0.05, minGSSize=5)
#result(res2)
## Reactome annotation system
#res3 <- tsea_mabs(drugs=drugs10, type="Reactome", pvalueCutoff=1)
#result(res3)