INGOR
|
CompFilter compares two networks.
This filter compares the input network with another network and pass the input network to the next filter without any modification.
This is originally designed to perform the comparison of the estimated network from the simulated data with the true structure. Therefore, the comparison result is shown by the sensitivity and the specificity, that are popular measures of evaluating the performance of machine learning algorithms.
The comparison result is output to the standard error (log file). The lines beginning with "COML:
" and "COMP:
" are the main part of the result. They are tab-delimited text. The line COML:
shows the column label showing the meanings of the tab-delimited columns. The line COMP:
shows values of the comparison result.
TP
: True positive edges. The number of edges that exist both in the input network and the second network (network read from the file).RTP
: Reversed positive edges. The number of edges that exist both in the input and second networks but the edge directions are different.FP
: False positive edges. The number of edges that exist in the input network but do not in the second network.FN
: False negative edges. The number of edges that do not exist in the input network but exist in the second network.TN
: True negative edges. The number of edge that do not exist both in the input and second networks.Sn
: Sensitivity = TP
/ (TP
+ FN
).Sp
: Specificty = TP
/ (TP
+ RTP
+ FP
).Snr
: Sensitivity allowing reverse edges = (TP
+ RTP
) / (TP
+ RTP
+ FN
).Spr
: Specifity allowing reverse edges = (TP
+ RTP
) / (TP
+ RTP
+ FP
).The comparison result can be output in network files. Use the tp
, fp
, and fn
options for saving true positive, false positive, and false negative edges in files.
file=
file File name of the network to read and compare.
type=
format Network file format type. See Network File Formats for available formats. If this is omitted, the extension of the file name given by file
is used to determine the format type.
args={
key [ =
values,
... ]}
Arguments for the network file format.
tp=
file tptype=
format tpargs={
key=
value,
...}
Arguments for the network file format used to save true positive edges.
fp=
file fptype=
format fpargs={
key=
value,
...}
Arguments for the network file format used to save false positive edges.
fn=
file fntype=
format fnargs={
key=
value,
...}
Arguments for the network file format used to save false negative edges.
output=
file Saves the comparison result in the specified file.
show_not_found
Outputs names of nodes that are not found in the true network.