This directory contains classes for working with pan-genomic graphs and frequented regions, based on the paper
Cleary, et al., "Exploring Frequented Regions in Pan-Genomic Graphs", IEEE/ACM Trans Comput Biol Bioinform. 2018 Aug 9. PMID:30106690 DOI:10.1109/TCBB.2018.2864564
This work was funded in part by the National Center for Genome Resources, Santa Fe, NM.
The project is set up with dependencies managed with the Gradle build tool. To build the distribution, simply run
$ ./gradlew installDist
This will create a distribution under build/install that is used by the various run scripts.
This contains two packages with similarly-named classes:
org.ncgr.pangenomics.allele which contains classes for working with allele-based sequence graphs org.ncgr.pangenomics.genotype which contains classes for working with genotype graphs
Basic graph-related classes, not particularly specific to frequented regions:
PangenomicGraph extends org.jgrapht.graph.DirectedAcyclicGraph and stores a graph with methods for reading it in from files and various output methods.
There is a main class for creating a graph from input data such as a GFA or VCF file.
Node encapsulates a node in a Graph: its ID (a long) and, for sequence graphs, its sequence.
NodeSet encapsulates a set of nodes in a Graph. NodeSet implements Comparable. There is a method merge() for merging two NodeSets.
(These are called "node clusters" in the paper above, but since I've implemented it as an extension of TreeSet, I've used "Set").
Path encapsulates a path through a Graph, along with its full sequence in the case of sequence graphs.
Frequented regions-related code.
FrequentedRegion stores a FrequentedRegion, containing a NodeSet along with the supporting subpaths of the full set of Paths in a Graph, and lots of methods.
FRFinder contains a main() method for finding FRs based on a bunch of parameters.
FRPair is a utility class that contains two FRs and the result of merging them, and is used in the search loop in FRFinder.findFRs().
LIBSVM-based Support Vector Machine classes.
Weka-based supervised classifier classes.