Has anybody looked at the predictive power of the 1000 Self-Organizing-Map states in ENCODE for extra TFBSs whose TFs haven't been chipped in ENCODE?

This is, given that the states described by the Self Organizing Map (SOM) in ENCODE where produced by a combination of histone marks and TF binding marks, and theoretically these 1000 states may represent underlying gene regulatory networks like the BMP example shown in the results, is there any predictive power to expand the regions with marks for extra TFBSs other than the ones that were used in ENCODE?

I can imagine how this can be tested by looking at publicly available ChIP-seq experiments done independently for other TFs in the same or similar cell line sets.