9 May 2015 in Epigenetics & Chromatin
DNA methylation is an epigenetic mark that balances plasticity with stability. While DNA methylation exhibits tissue specificity, it can also vary with age and potentially environmental exposures. In studies of DNA methylation, samples from specific tissues, especially brain, are frequently limited and so surrogate tissues are often used. As yet, we do not fully understand how DNA methylation profiles of these surrogate tissues relate to the profiles of the central tissue of interest. We have adapted principal component analysis to analyze data from the Illumina 450K Human Methylation array using a set of 17 individuals with 3 brain regions and whole blood. All of the top five principal components in our analysis were associated with a variable of interest: principal component 1 (PC1) differentiated brain from blood, PCs 2 and 3 were representative of tissue composition within brain and blood, respectively, and PCs 4 and 5 were associated with age of the individual (PC4 in brain and PC5 in both brain and blood). We validated our age-related PCs in four independent sample sets, including additional brain and blood samples and liver and buccal cells. Gene ontology analysis of all five PCs showed enrichment for processes that inform on the functions of each PC. Principal component analysis (PCA) allows simultaneous and independent analysis of tissue composition and other phenotypes of interest. We discovered an epigenetic signature of age that is not associated with cell type composition and required no correction for cellular heterogeneity.
8 April 2023 in ACM Transactions on Quantum Computing
Quantum processing units (QPUs) executing annealing algorithms have shown promise in optimization and simulation applications. Hybrid algorithms are a natural bridge to larger applications. We present a simple greedy method for solving larger-than-QPU lattice-structured Ising optimization problems. The method, implemented in the open source D-Wave Hybrid framework, uses a QPU coprocessor operating with generic parameters. Performance is evaluated for standard spin-glass problems on two lattice types with up to 11,616 spin variables, double the size that is directly programmable on any available QPU. The proposed method is shown to converge to low-energy solutions faster than an open source simulated annealing method that is either directly employed or substituted as a coprocessor in the hybrid method. Using newer Advantage QPUs in place of D-Wave 2000Q QPUs is shown to enhance convergence of the hybrid method to low energies and to achieve a lower final energy.
16 December 2023 in ACM Transactions on Quantum Computing
We introduce quantum utility , a new approach to evaluating quantum performance that aims to capture the user experience by considering the overhead costs associated with a quantum computation. A demonstration of quantum utility by the quantum processing unit (QPU) shows that the QPU can outperform classical solvers at some tasks of interest to practitioners, when considering the costs of computational overheads. A milestone is a test of quantum utility that is restricted to a specific subset of overhead costs and input types. We illustrate this approach with a benchmark study of a D-Wave annealing-based QPU versus seven classical solvers for a variety of problems in heuristic optimization. We consider overhead costs that arise in standalone use of the D-Wave QPU (as opposed to a hybrid computation). We define three early milestones on the path to broad-scale quantum utility. Milestone 0 is the purely quantum computation with no overhead costs and is demonstrated implicitly by positive results on other milestones. We evaluate the performance of a D-Wave Advantage QPU with respect to milestones 1 and 2: For milestone 1, the QPU outperformed all classical solvers in 99% of our tests. For milestone 2, the QPU outperformed all classical solvers in 19% of our tests, and the scenarios in which the QPU found success correspond to cases where classical solvers most frequently failed. This approach of isolating subsets of overheads for separate analysis reveals distinct mechanisms in quantum versus classical performance, which explain the observed differences in patterns of success and failure. We present evidence-based arguments that these distinctions bode well for annealing quantum processors to support demonstrations of quantum utility on ever-expanding classes of inputs and with more challenging milestones in the very near future.
11 October 2018 in BMC Bioinformatics
DNA inside eukaryotic cells wraps around histones to form the 11nm chromatin fiber that can further fold into higher-order DNA loops, which may depend on the binding of architectural factors. Predicting how the DNA will fold given a distribution of bound factors, here viewed as a type of sequence, is currently an unsolved problem and several heterogeneous polymer models have shown that many features of the measured structure can be reproduced from simulations. However a model that determines the optimal connection between sequence and structure and that can rapidly assess the effects of varying either one is still lacking. Here we train a dense neural network to solve for the local folding of chromatin, connecting structure, represented as a contact map, to a sequence of bound chromatin factors. The network includes a convolutional filter that compresses the large number of bound chromatin factors into a single 1D sequence representation that is optimized for predicting structure. We also train a network to solve the inverse problem, namely given only structural information in the form of a contact map, predict the likely sequence of chromatin states that generated it. By carrying out sensitivity analysis on both networks, we are able to highlight the importance of chromatin contexts and neighborhoods for regulating long-range contacts, along with critical alterations that affect contact formation. Our analysis shows that the networks have learned physical insights that are informative and intuitive about this complex polymer problem.
5 February 2018 in PLOS Computational Biology
The packaging of DNA inside a nucleus shows complex structure stabilized by a host of DNA-bound factors. Both the distribution of these factors and the contacts between different genomic locations of the DNA can now be measured on a genome-wide scale. This has advanced the development of models aimed at predicting the conformation of DNA given only the locations of bound factors—the chromatin folding problem. Here we present a maximum-entropy model that is able to predict a contact map representation of structure given a sequence of bound factors. Non-local effects due to the sequence neighborhood around contacting sites are found to be important for making accurate predictions. Lastly, we show that the model can be used to infer a sequence of bound factors given only a measurement of structure. This opens up the possibility for efficiently predicting sequence regions that may play a role in generating cell-type specific structural differences. The three-dimensional folding of DNA inside the nucleus into specific conformations is necessary for the proper functioning of cells. These structures can be measured by chromosome conformation capture methods (Hi-C) that report the number of times that each pair of genomic sites are found in proximal location in a cell population experiment. A number of protein complexes that bind to the DNA have been discovered to be responsible for the stabilization of such conformations. However, identifying the precise relation between the positioning of binding proteins and the resulting structures is still an open problem. Here we present a maximum-entropy method able to predict Hi-C contact probabilities from a sequence of binding factors without the need of performing any polymer simulations. We envision that this method will allow experimentalists to efficiently calculate the expected structural effect of altering the sequence of binding factors. In addition, we also show that our model is capable of solving the inverse problem, namely predicting the underlying sequence of binding factors from a set of observed contact probabilities.