Triplex were predicted using Bioconductor Triplex package.

The low stringency file contains all triplexes detected by package.

The high stringency only the Triplex that had a p=value < 0.05.
These p-value is caculated for overlapping triplexes and helps select the triplex most likely to form 
in any given genomic region (when there are overlapping triplex). (see Bioconductor '
Triplex package and reference

 

A dynamic programming algorithm for identification of triplex-forming sequences.
Lexa M, Martínek T, Burgetová I, Kopecek D, Brázdová M.
Bioinformatics. 2011 Sep 15;27(18):2510-7.