Avoiding False Positive Conclusions in Molecular Simulation: The Importance of Replicas

Abstract

Molecular simulations are a computational technique used to investigate the dynamics of proteins and other molecules. The free energy landscape of these simulations is often rugged, and minor differences in the initial velocities, floating-point precision, or underlying hardware can cause identical simulations (replicas) to take different paths in the landscape. In this study we investigated the magnitude of these effects based on 310 000 ns of simulation time. We performed 100 identically parametrized replicas of 3000 ns each for a small 10 amino acid system as well as 100 identically parametrized replicas of 100 ns each for an 827 residue T-cell receptor/MHC system. Comparing randomly chosen subgroups within these replica sets, we estimated the reproducibility and reliability that can be achieved by a given number of replicas at a given simulation time. These results demonstrate that conclusions drawn from single simulations are often not reproducible and that conclusions drawn from multiple shorter replicas are more reliable than those from a single longer simulation. The actual number of replicas needed will always depend on the question asked and the level of reliability sought. On the basis of our data, it appears that a good rule of thumb is to perform a minimum of five to 10 replicas.

Comment on ‘Response to comment on ‘Valid molecular dynamics simulations of human hemoglobin require a surprisingly large box size’’

“…We argue that, as molecular dynamics simulations behave as chaotic systems and therefore sample stochastically, a single simulation replica per simulation box size in which a transition is observed (or not) is not sufficient to support conclusions on box size effects on that transition. Rather, N=1 evidence is anecdotal in nature, as had been pointed out by one of the reviewers of the original paper (1) and in line with a recent investigation into the importance of mutiple replicas in MD simulations (4). Indeed, in our investigation, in which we applied an order of magnitude more statistics (N=10-20) (2), we found a large scatter of transition times for each studied box size.…”

How quickly can we predict trimethoprim resistance using alchemical free energy methods?

“…Such methods could be used to screen out mutations that have no effect on an antibiotic, leaving only the marginal cases for computationally more intensive approaches such as proposed here. As computational resource becomes faster and more widespread, the broader field of molecular simulation is gradually moving away from running single simulations towards running large number of replicas [28,29]. This potentially exposes underlying drawbacks with the molecular dynamics codes and how we, as computational scientists, typically work.…”

PyRod Enables Rational Homology Model‐based Virtual Screening Against MCHR1

“…The simulation time of the 10 replicates was increased from 10 ns to 30 ns compared to the original PyRod publication to relax artifacts introduced through homology modeling . The last 10 ns of each replicate were analyzed with PyRod granting sufficient sampling of side chain conformations …”