Create a Consensus Adjacency Matrix from Multiple Networks

Builds a consensus adjacency matrix from a list of networks using one of three methods: "vote", "union", or "INet".

Usage

create_consensus(
  adj_matrix_list,
  method = "vote",
  weighted_list = NULL,
  theta = 0.04,
  threshold = 0.5,
  ncores = 1,
  tolerance = 0.1,
  nitermax = 50,
  verbose = FALSE
)

Arguments

adj_matrix_list

A list of binary adjacency matrices (square, 0/1) with identical dimensions and matching row/column names.

method

Character string specifying the consensus strategy. One of:

"vote" (default): An edge is included if supported by at least threshold fraction of matrices.
"union": An edge is included if present in any matrix.
"INet": Combines normalized weighted matrices using consensusNet.

weighted_list

A list of weighted adjacency matrices (required if method = "INet").

theta

Numeric. Tuning parameter passed to consensusNet (default: 0.04).

threshold

Numeric between 0 and 1. Threshold for "vote" and "INet" methods. Default is 0.5.

ncores

Integer. Number of CPU cores to use when method = "INet". Default is 1.

tolerance

Numeric. Tolerance for differences between similar graphs in INet method. Default is 0.1.

nitermax

Integer. Maximum number of iterations for INet algorithm. Default is 50.

verbose

Logical. If TRUE, display verbose output for INet method. Default is FALSE.

Value

A square consensus adjacency matrix (binary or weighted, depending on the method).

Details

Consensus construction depends on the selected method:

vote: Counts the presence of each edge across all matrices and includes edges supported by at least threshold × N matrices.
union: Includes any edge that appears in any matrix.
INet: Multiplies binary matrices by corresponding weighted matrices, normalizes the results, and applies consensusNet to generate a consensus network.

For "INet", both binary and weighted adjacency matrices must be provided with matching dimensions.

Examples

data(count_matrices)

networks <- infer_networks(
    count_matrices_list = count_matrices,
    method = "GENIE3",
    nCores = 1
)
head(networks[[1]])
#>   regulatoryGene targetGene    weight
#> 1          ARPC2      ARPC3 0.2002341
#> 2          HLA-A       CD74 0.1942305
#> 3           CD3E       CD3D 0.1647460
#> 4          ARPC3      ARPC2 0.1610786
#> 5           CD3E        JUN 0.1571037
#> 6          HLA-E        FOS 0.1500092

wadj_list <- generate_adjacency(networks)
swadj_list <- symmetrize(wadj_list, weight_function = "mean")

binary_listj <- cutoff_adjacency(
    count_matrices = count_matrices,
    weighted_adjm_list = swadj_list,
    n = 2,
    method = "GENIE3",
    quantile_threshold = 0.99,
    nCores = 1,
    debug = TRUE
)
#> [Method: GENIE3] Matrix 1 → Cutoff = 0.10286
#> [Method: GENIE3] Matrix 2 → Cutoff = 0.10876
#> [Method: GENIE3] Matrix 3 → Cutoff = 0.10636
head(binary_listj[[1]])
#>       ACTG1 ARPC2 ARPC3 BTF3 CD3D CD3E CD74 CFL1 COX4I1 COX7C CXCR4 EEF1A1
#> ACTG1     0     0     0    0    0    0    0    0      0     0     0      0
#> ARPC2     0     0     1    0    0    0    0    0      0     0     0      0
#> ARPC3     0     1     0    0    0    0    0    0      0     0     0      0
#> BTF3      0     0     0    0    0    0    0    0      0     0     0      0
#> CD3D      0     0     0    0    0    1    0    0      0     0     0      0
#> CD3E      0     0     0    0    1    0    0    0      0     0     0      0
#>       EEF1D EEF2 EIF1 EIF3K EIF4A2 FOS FTH1 FTL GNB2L1 HLA-A HLA-B HLA-C HLA-E
#> ACTG1     0    0    0     0      0   0    0   0      0     0     0     0     0
#> ARPC2     0    0    0     0      0   0    0   0      0     0     0     0     1
#> ARPC3     0    0    0     0      0   0    0   0      0     0     0     0     0
#> BTF3      0    0    0     0      0   0    0   0      0     0     0     0     0
#> CD3D      0    0    0     0      0   0    0   0      0     0     0     0     0
#> CD3E      0    0    0     0      0   0    0   0      0     0     0     0     0
#>       JUN JUNB MYL12B MYL6 NACA PABPC1 PFN1 TMSB4X UBA52 UBC
#> ACTG1   0    0      0    0    0      0    0      0     0   0
#> ARPC2   0    0      0    0    0      0    0      0     0   0
#> ARPC3   0    0      0    0    0      0    0      0     0   0
#> BTF3    0    0      0    0    0      0    0      0     0   0
#> CD3D    0    0      0    0    0      0    0      0     0   0
#> CD3E    0    0      0    0    0      0    0      0     0   0

consensus <- create_consensus(binary_listj, method = "vote")
head(consensus)
#>       ACTG1 ARPC2 ARPC3 BTF3 CD3D CD3E CD74 CFL1 COX4I1 COX7C CXCR4 EEF1A1
#> ACTG1     0     0     0    0    0    0    0    0      0     0     0      0
#> ARPC2     0     0     1    0    0    0    0    0      0     0     0      0
#> ARPC3     0     1     0    0    0    0    0    0      0     0     0      0
#> BTF3      0     0     0    0    0    0    0    0      0     0     0      0
#> CD3D      0     0     0    0    0    0    0    0      0     0     0      0
#> CD3E      0     0     0    0    0    0    0    0      0     0     0      0
#>       EEF1D EEF2 EIF1 EIF3K EIF4A2 FOS FTH1 FTL GNB2L1 HLA-A HLA-B HLA-C HLA-E
#> ACTG1     0    0    0     0      0   0    0   0      0     0     0     0     0
#> ARPC2     0    0    0     0      0   0    0   0      0     0     0     0     0
#> ARPC3     0    0    0     0      0   0    0   0      0     0     0     0     0
#> BTF3      0    0    0     0      0   0    0   0      0     0     0     0     0
#> CD3D      0    0    0     0      0   0    0   0      0     0     0     0     0
#> CD3E      0    0    0     0      0   0    0   0      0     0     0     0     0
#>       JUN JUNB MYL12B MYL6 NACA PABPC1 PFN1 TMSB4X UBA52 UBC
#> ACTG1   0    0      0    0    0      0    0      0     0   0
#> ARPC2   0    0      0    0    0      0    0      0     0   0
#> ARPC3   0    0      0    0    0      0    0      0     0   0
#> BTF3    0    0      0    0    0      0    0      0     0   0
#> CD3D    0    0      0    0    0      0    0      0     0   0
#> CD3E    0    0      0    0    0      0    0      0     0   0