MOTIVATION: Human immunodeficiency virus (HIV) drug resistance is a global healthcare issue. The emergence of drug resistance influenced the efficacy of treatment regimens, thus stressing the importance of treatment adaptation. Computational methods predicting the drug resistance profile from genomic data of HIV isolates are advantageous for monitoring drug resistance in patients. However, existing computational methods for drug resistance prediction are either not suitable for emerging HIV strains with complex mutational patterns or lack interpretability, which is of paramount importance in clinical practice. The approach reported here overcomes these limitations and combines high accuracy of predictions and interpretability of the models. RESULTS: In this work, a new methodology based on generative topographic mapping (GTM) for biological sequence space representation and quantitative genotype-phenotype relationships prediction purposes was introduced. The GTM-based resistance landscapes allowed us to predict the resistance of HIV strains based on sequencing and drug resistance data for three viral proteins [integrase (IN), protease (PR) and reverse transcriptase (RT)] from Stanford HIV drug resistance database. The average balanced accuracy for PR inhibitors was 0.89 ± 0.01, for IN inhibitors 0.85 ± 0.01, for non-nucleoside RT inhibitors 0.73 ± 0.01 and for nucleoside RT inhibitors 0.84 ± 0.01. We have demonstrated in several case studies that GTM-based resistance landscapes are useful for visualization and analysis of sequence space as well as for treatment optimization purposes. Here, GTMs were applied for the in-depth analysis of the relationships between mutation pattern and drug resistance using mutation landscapes. This allowed us to predict retrospectively the importance of the presence of particular mutations (e.g. V32I, L10F and L33F in HIV PR) for the resistance development. This study highlights some perspectives of GTM applications in clinical informatics and particularly in the field of sequence space exploration. AVAILABILITY AND IMPLEMENTATION: https://github.com/karinapikalyova/ISIDASeq. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Lattice-bound folding models are often used by computer scientists as a simplified instance of the chain folding problem (featuring, at the other end of the complexity spectrum, conformational sampling of biologically important molecules). Lattice-bound folding is relatively fast, thus well suited for benchmarking of nature-inspired optimization heuristics. Yet, it is not clear whether the benchmark-winning heuristics are necessarily the best suited for the more complex problems of atom-level resolution conformational sampling. This paper reports the design of - as far as possible - equivalent evolutionary operators and algorithms, for both chains and molecules, to explicitly check whether chain folding may serve as “training ground” for real sampling protocols.

Conformational sampling, the computational prediction of the experimental geometries of small proteins (folding) or of protein-ligand complexes (docking), is often cited as one of the most challenging multimodal optimization problems. Due to the extreme ruggedness of the energy landscape as a function of geometry, sampling heuristics must rely on an appropriate trade-off between global and local searching efforts. A previously reported ldquoplanetary strategyrdquo, a generalization of the classical island model used to deploy a hybrid genetic algorithm on computer grids, has shown a good ability to quickly discover low-energy geometries of small proteins and sugars, and sometimes even pinpoint their native structures-although not reproducibly. The procedure focused on broad exploration and used a tabu strategy to avoid revisiting the neighborhood of known solutions, at the risk of ldquoburyingrdquo important minima in overhastily set tabu areas. The strategy reported here, termed ldquodivide-and-conquer planetary modelrdquo couples this global search procedure to a local search tool. Grid nodes are now shared between global and local exploration tasks. The phase space is cut into ldquocellsrdquo corresponding to a specified sampling width for each of the N degrees of freedom. Global search locates cells containing low-energy geometries. Local searches pinpoint even deeper minima within a cell. Sampling width controls the important trade-off between the number of cells and the local search effort needed to reproducibly sample each cell. The probability to submit a cell to local search depends on the energy of the most stable geometry found within. Local searches are allotted limited resources and are not expected to converge. However, as long as they manage to discover some deeper local minima, the explored cell remains eligible for further local search, now relying on the improved energy level to enhance chances to be picked again. This competition prevents the system to waste too much effort in fruitless local searches. Eventually, after a limited number of local searches, a cell will be ldquoclosedrdquo and used - first as ldquoseedrdquo, later as tabu zone-to bias future global searches. Technical details and some folding and docking results will be discussed.

Molecular docking is an essential tool for drug design. It helps the scientist to rapidly know if two molecules, respectively called ligand and receptor, can be combined together to obtain a stable complex. We propose a new multi-objective model combining an energy term and a surface term to gain such complexes. The aim of our model is to provide complexes with a low energy and low surface. This model has been validated with two multi-objective genetic algorithms on instances from the literature dedicated to the docking benchmarking.

Computational simulations of conformational sampling in general, and of macromolecular folding in particular represent one of the most important and yet one of the most challenging applications of computer science in biology and medicinal chemistry. The advent of GRID computing may trigger some major progress in this field. This paper presents our first attempts to design GRID-based conformational sampling strategies, exploring the extremely rugged energy response surface in function of molecular geometry, in search of low energy zones through phase spaces of hundreds of degrees of freedom. We have generalized the classical island model deployment of genetic algorithms (GA) to a "planetary" model where each node of the grid is assimilated to a "planet" harboring quasi-independent multi-island simulations based on a hybrid GA-driven sampling approach. Although different "planets" do not communicate to each other-thus minimizing inter-CPU exchanges on the GRID-each new simulation will benefit from the preliminary knowledge extracted from the centralized pool of already visited geometries, located on the dispatcher machine, and which is disseminated to any new "planet". This "panspermic" strategy allows new simulations to be conducted such as to either be attracted towards an apparently promising phase space zone (biasing strategies, intensification procedures) or to avoid already in-depth sampled (tabu) areas. Successful folding of mini-proteins typically used in benchmarks for all- atoms protein simulations has been observed, although the reproducibility of these highly stochastic simulations in huge problem spaces is still in need of improvement. Work on two structured peptides (the "tryptophane cage" 1L2Y and the "tryptophane zipper" 1LE1) used as benchmarks for all-atom protein folding simulations has shown that the planetary model is able to reproducibly sample conformers from the neighborhood of the native geometries. However, within these neighborhoods (within ensembles of conformers similar to models published on hand of experimental geometry determinations), the energy landscapes are still extremely rugged. Therefore, simulations in general produce "correct" geometries (similar enough to experimental model for any practical purposes) which sometimes unfortunately correspond to relatively high energy levels and therefore are less stable than the most stable among misfolded conformers. The method thus reproducibly visits the native phase space zone, but fails to reproducibly hit the bottom of its rugged energy well. Intensifications of local sampling may in principle solve this problematic behavior, but is limited by computational resources. The quest for the optimal time point at which a phase space zone should stop being intensively searched and declared tabu, a very difficult problem, is still awaiting for a practically useful solution.