EdgeContribFilter (--ec) calculates various contributions of edges with respect to a sample or a set of samples.

This filter does not change the network structure. The calculated values are attached as edge attributes.

The output argument can be used to output an ECv matrix as a text file. If the input data does not contain all the nodes in the network, edges with such nodes are not written in the file.

Arguments

data=file

Input data file in GDF format. The file needs to have the same variables (attributes) as in the network.

data_args={ key=value, ... }

Arguments for the input data. See document of ytGDF_read_fp() for available arguments.

method= ( RC | RCr | ECv | RC2 | ECv2)

Method (metric) of contributions. RC: relative contribution, RCr: Reletive contribution ratio, ECv: edge contribution value, RC2: Same as RC except that -1 is set for single parent nodes. ECv2: Same as ECv except that NA (NaN) values are ignored and output as NA values. (Default: RC)

mode= ( pid | each )

Calculation mode. pid: summarizing samples for the same primary IDs, each: calculates for each sample. (default: pid)

output=file

out=file

Output file name. If specified, the calculated ECv matrix is output in the specified file as a tab separated text. Each row in the file represents an edge. The first two columns are the parent and the child of an edge, followed by the caluclated values. The order of the values is the same as in the input data file. The first line in the file is the header row.

dynamic

d

Dynamic model

sdhybrid

sd

Static-dynamic hybrid model.

sampleid

Suppresses to add sample IDs to labeles of the calculated edge contribution values.

fit

Re-fit the model to the data instead of using the model parameters in the network.

score=str

score function name.

score_args={key=value,...}

Arguments for the score function.

name

Specifies to output the "name" column which is a "Parent_Child" style name of an edge in the output file, where Parent and Child represents the names of the parent and child of the edge, respectively. The binding character ("_" by default) can be changed by the binder argument.

binder=str

The binding character used to output the "name" column. "_" (underscore) is used by default.

edgeprop= ( on | off )

Specifies whether or not calculated values are stored in the network as edge properties. By default, "on" is assumed.

nan=str

na=str

The string used to output NaN (not a number), NA (non available), or missing values. (Default: "nan")

MPI support

MPI parallelization supported. Parallelization is based on distributing the child nodes among processes. The calculated results are gathered in the root process, and only the root process outputs the results into a file if "output" option is specified.

Filters | INGOR Manual