Triplex Domain Finder (TDF) uses Triplexator to find DNA binding sites (DBS) and RNA binding sites (RBS) of a lncRNA in DNA regions. One set of regions comprises potential lncRNA targets (target regions), e.g. promoters of genes differentially expressed after FENDRR knockdown. A second set defined as non-target regions comprises promoters of genes not differentially expressed. Next, we list all candidate DNA binding domains (DBDs) by finding regions in the RNA with overlapping RBS (see figure below). We only consider candidate DBDs with more than k DBS in target regions (k=20 as default). Finally, we test whether the number of target regions with at least one DBS is higher than the number of non-target regions with at least one DBS for a given DBD.
The promoter test evaluates the triple helix binding potential of a lncRNA in the promoters of candidate genes, e.g. genes differentially expressed in a particular functional study. For this, it compares binding statistics on the promoters of candidate genes (target regions) with the binding statistics in all other promoters (non-target regions).
Genomic Region Test
The genomic region test evaluates the triple helix binding potential of a lncRNA in a set of regions in the DNA, as indicated by genome-wide essays measuring the interaction of lncRNAs with DNA (ChIRP-Seq and CHART-Seq). Due to lack of a non-target region set, we select random genomic regions as non-target regions and use an empirical statistical test. This test evaluates whether the number of target regions with at least one DBS is larger than the number of non-target regions with at least one DBS.
Exact details are provided here:
S. Hanzelmann, C.C. Kuo, M. Kalwa, W. Wagner, I. G. Costa, Triplex Domain Finder: Detection of Triple Helix Binding Domains in Long Non-Coding RNAs, biorxiv, 2016 [paper].