Viruses are known for their extremely high mutation rates,
allowing
them to evade both the human immune system and many forms of standard
medicine. Despite this, the RNA dependent RNA polymerase (RdRp) of
the RNA viruses has been largely conserved, and any significant mutation
of this protein is unlikely. The recent COVID-19 pandemic presents
a need for therapeutics. We have designed a de novo drug design algorithm
that generates strong binding ligands from scratch, based on only
the structure of the target protein’s receptor. In this paper,
we applied our method to target SARS-CoV-2 RdRp and generated several
de novo molecules. We then chose some drug molecules based on the
structural similarity to some of our strongest binding de novo molecules.
Subsequently, we showed, using rigorous all-atom explicit-water free
energy calculations in near-microsecond time scales using state-of-the-art
well-tempered metadynamics simulations, that some of our de novo generated
ligands bind more strongly to RdRp than the recent FDA approved drug
remdesivir in its active form, remdesivir triphosphate (RTP). We elucidated
the binding mechanism for some of the top binders and compared it
with RTP. We believe that this work will be useful both by presenting
lead structures for RdRp inhibition and by delivering key insights
into the residues of the protein potentially involved in the binding/unbinding
of these small molecule drugs, leading to more targeted studies in
the future.