To determine the probability that an edge is "different" between the networks, we first subtracted the z-score weight values estimated by PANDA for the two networks and then determined the value of the inverse cumulative distribution for this difference. The product of these two probabilities represents the probability than an edge is both "supported" and "different." We select edges for which this combined probability is greater than a threshold probability (default value is 0.8).
panda.diff.edges( panda.net1, panda.net2, threshold = 0.8, condition_name = "cond.1" )
| panda.net1 | vector indicating the PANDA networks of one condition or phenotype. |
|---|---|
| panda.net2 | vector indicating the PANDA networks of another compared condition orphenotype. |
| threshold | numerical vector indicating a threshold probability to select select edges. |
| condition_name | string vector indicating the condition name of net1 |
a data.frame with five columns: tf, gene, motif, Score and defined condition name(the row with "T" in this column means this egde belongs to first condition or phenotype, "F" means this edge belongs to the second condition or phenotype)
# refer to four input datasets files in inst/extdat treated_expression_file_path <- system.file("extdata", "expr4_matched.txt", package = "netZooR", mustWork = TRUE) control_expression_file_path <- system.file("extdata", "expr10_matched.txt", package = "netZooR", mustWork = TRUE) motif_file_path <- system.file("extdata", "chip_matched.txt", package = "netZooR", mustWork = TRUE) ppi_file_path <- system.file("extdata", "ppi_matched.txt", package = "netZooR", mustWork = TRUE) # Run PANDA for treated and control network treated_all_panda_result <- panda.py(expr_file = treated_expression_file_path, motif_file= motif_file_path, ppi_file = ppi_file_path, modeProcess="legacy", remove_missing = TRUE ) control_all_panda_result <- panda.py(expr_file = control_expression_file_path, motif_file= motif_file_path, ppi_file= ppi_file_path, modeProcess="legacy", remove_missing = TRUE ) # access PANDA regulatory network treated_net <- treated_all_panda_result$panda control_net <- control_all_panda_result$panda merged.panda <- panda.diff.edges(treated_net, control_net, condition_name="treated")