English: Pool-based ensemble modeling (left) starts by generating a pool of random or semi-random conformations based on the protein sequence. Subsets of conformations are selected iteratively from the pool and theoretical parameters are calculated for each conformer in the subset. The final ensemble consists of conformations for which the theoretical parameters are in agreement with the experimental data. MD-based approaches start by initiating short replica MD simulations in parallel using an initial conformation. The MD replicas are constrained with the experimental data. The final ensemble is a combination of the resulting replica runs.