A new entropy model for RNA: part IV, The Minimum Free Energy (mFE) and the thermodynamically most-probable folding pathway (TMPFP)

Abstract

Here we discuss four important questions (1) how can we be sure that the thermodynamically most-probable folding-pathway yields the minimum free energy for secondary structure using the dynamic programming algorithm (DPA) approach, (2) what are its limitations, (3) how can we extend the DPA to find the minimum free energy with pseudoknots, and finally (4) what limitations can we expect to find in a DPA approach for pseudoknots. It is our supposition that some structures cannot be fit uniquely by the DPA, but may exist in real biology situations when disordered regions in the biomolecule are necessary. These regions would be identifiable by using suboptimal structure analysis. This grants us some qualitative tools to identify truly random RNA sequences, because such are likely to have greater degeneracy in their thermodynamically most-probable folding-pathway.

Free energy based high-resolution modeling of CTCF-mediated chromatin loops for human genome

“…Hence, a result of this optimization procedure is the generation of meta-structures (ranked with respect to free energy) that express the pair-wise interactions of the chromatin chain as well as multi-chain interactions. The method has been demonstrated to work for other problems such as RNA structure prediction with pseudoknots (Dawson et al, 2007;Dawson et al, 2014) and in proof of principle in protein structure prediction (Dawson et al, 2005). This is the first time to test this GPC-based entropy model based on chromatin structures.…”

“…The experimentally observed pair interaction frequency (PIF) can be used to estimate the binding affinity or statistical weight for a given pair. A pair binding free energy based polymer model can be constructed to model motifs from a statistical potential using the PIFs (from ChIA-PET data) and entropy loss due to folding model that measures the entropy loss due to the formation of contacts, or formation of cross links ((Dawson et al, 2001a, b;Dawson et al, 2014;Dawson and Kawai, 2015) and references therein). This would inform us of the optimal structures of these loops, or collections of loops from chromatin capture domains (CCDs) and whether such a distribution is largely flexible and shifting between multiple structures (a likely case for active regions; e.g., related to the presence of RNA Pol II), or a single structure that is largely frozen (a likely case for an inactive region of the genome, enriched with insulators such as CTCF proteins).…”