evaluation.networkperformance
Class ROC
java.lang.Object
evaluation.networkperformance.ROC
public class ROC
- extends Object
Contains static methods to calculate ROC curves for protein networks. These
ROC curves are a quality measure and describe how good the scores within the
network fit to a given complex set.
Calculation procedure
ROC curves are computed as follows.
- As there are too many possible negative edges a negative set is sampled.
To generate this set, edges between random proteins are choosen which are
not within a complex and optionally not colocalized (which of requires course
requires
LocalizationData
).
- Each edge in the network is assigned one of the following states:
- true positive if the edge is contained in the complex set
- false positive if the edge is contained in the negative set
(see above)
- unknown otherwise
-
[text mit threshold runterziehen und so]
- Author:
- jan
Constructor Summary |
ROC()
|
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
ROC
public ROC()
calculateROCCurves
public static List<ROCCurve> calculateROCCurves(List<ProteinNetwork> scoreNetworks,
ComplexSet reference,
ComplexSet referenceForNegativeSet,
LocalizationData locData)
- Calculate the ROC curves for a given set of networks. Can use a different
complex set for the identification
- Parameters:
scoreNetworks
- list of networks for which the ROC curves will be calculatedreference
- reference set to be usedreferenceForNegativeSet
- reference setlocData
- localization data
, use null
if you do not want to use localization data for negative set
generation
- Returns:
- list of ROC curves for the given networks