public class GeneSetMapTools extends Object
Note that in ErmineJ 3 these are no longer used.
Modifier and Type | Field and Description |
---|---|
protected static org.apache.commons.logging.Log |
log |
Constructor and Description |
---|
GeneSetMapTools() |
Modifier and Type | Method and Description |
---|---|
static Collection<String> |
getSimilarities(String classId,
Map<String,Collection<String>> classesToSimilarMap) |
static void |
ignoreSimilar(double fractionSameThreshold,
GeneAnnotations ga,
StatusViewer messenger,
int maxClassSize,
int minClassSize,
double bigClassPenalty)
Remove classes which are too similar to some other class.
|
static void |
removeAspect(GeneAnnotations ga,
GeneSetTerms gon,
StatusViewer messenger,
String aspect) |
static void |
removeBySize(GeneAnnotations ga,
StatusViewer messenger,
int minClassSize,
int maxClassSize)
Remove gene sets that don't meet certain criteria.
|
public static Collection<String> getSimilarities(String classId, Map<String,Collection<String>> classesToSimilarMap)
classId
- classesToSimilarMap
- public static void ignoreSimilar(double fractionSameThreshold, GeneAnnotations ga, StatusViewer messenger, int maxClassSize, int minClassSize, double bigClassPenalty)
Remove classes which are too similar to some other class. In addition, the user can select a penalty for large gene sets. Thus when two gene sets are found to be similar, the decision of which one to keep can be tuned based on the size penalty. We find it useful to penalize large gene sets so we tend to keep smaller ones (but not too small). Useful values of the penalty are above 1 (a value of 1 will result in the larger class always being retained).
The amount of similarity to be tolerated is set by the parameter fractionSameThreshold, representing the fraction of genes in the smaller class which are also found in the larger class. Thus, setting this threshold to be 0.0 means that no overlap is tolerated. Setting it to 1 means that classes will never be discarded.
fractionSameThreshold
- A value between 0 and 1, indicating how similar a class must be before it gets
ditched.ga
- messenger
- For updating a log.maxClassSize
- Large class considered. (that doesn't mean they are removed)minClassSize
- Smallest class considered. (that doesn't mean they are removed)bigClassPenalty
- A value greater or equal to one, indicating the cost of retaining a larger class in favor
of a smaller one. The penalty is scaled with the difference in sizes of the two classes being considered,
so very large classes are more heavily penalized.public static void removeAspect(GeneAnnotations ga, GeneSetTerms gon, StatusViewer messenger, String aspect)
ga
- gon
- messenger
- aspect
- public static void removeBySize(GeneAnnotations ga, StatusViewer messenger, int minClassSize, int maxClassSize)
ga
- messenger
- minClassSize
- maxClassSize
- Copyright © 2002-2013 UBC Centre for High-Throughput Biology. All Rights Reserved.