Using Differential Evolution to Design Optimal Experiments.
Ontology highlight
ABSTRACT:
SUBMITTER: Stokes Z
PROVIDER: S-EPMC7088454 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:We propose a new method for optimal experimental design of population pharmacometric experiments based on global search methods using interval analysis; all variables and parameters are represented as intervals rather than real numbers. The evaluation of a specific design is based on multiple simulations and parameter estimations. The method requires no prior point estimates for the parameters, since the parameters can incorporate any level of uncertainty. In this respect, it is similar to robust optimal design. Representing sampling times and covariates like doses by intervals gives a direct way of optimizing with rigorous sampling and dose intervals that can be useful in clinical practice. Furthermore, the method works on underdetermined problems for which traditional methods typically fail.
| S-EPMC3231860 | biostudies-other
Project description:Modern biological experiments are becoming increasingly complex, and designing these experiments to yield the greatest possible quantitative insight is an open challenge. Increasingly, computational models of complex stochastic biological systems are being used to understand and predict biological behaviors or to infer biological parameters. Such quantitative analyses can also help to improve experiment designs for particular goals, such as to learn more about specific model mechanisms or to reduce prediction errors in certain situations. A classic approach to experiment design is to use the Fisher information matrix (FIM), which quantifies the expected information a particular experiment will reveal about model parameters. The Finite State Projection based FIM (FSP-FIM) was recently developed to compute the FIM for discrete stochastic gene regulatory systems, whose complex response distributions do not satisfy standard assumptions of Gaussian variations. In this work, we develop the FSP-FIM analysis for a stochastic model of stress response genes in S. cerevisae under time-varying MAPK induction. We verify this FSP-FIM analysis and use it to optimize the number of cells that should be quantified at particular times to learn as much as possible about the model parameters. We then extend the FSP-FIM approach to explore how different measurement times or genetic modifications help to minimize uncertainty in the sensing of extracellular environments, and we experimentally validate the FSP-FIM to rank single-cell experiments for their abilities to minimize estimation uncertainty of NaCl concentrations during yeast osmotic shock. This work demonstrates the potential of quantitative models to not only make sense of modern biological data sets, but to close the loop between quantitative modeling and experimental data collection.
| S-EPMC7515449 | biostudies-literature
Project description:MotivationWe addressed the problem of inferring gene regulatory network (GRN) from gene expression data of knockout (KO) experiments. This inference is known to be underdetermined and the GRN is not identifiable from data. Past studies have shown that suboptimal design of experiments (DOE) contributes significantly to the identifiability issue of biological networks, including GRNs. However, optimizing DOE has received much less attention than developing methods for GRN inference.ResultsWe developed REDuction of UnCertain Edges (REDUCE) algorithm for finding the optimal gene KO experiment for inferring directed graphs (digraphs) of GRNs. REDUCE employed ensemble inference to define uncertain gene interactions that could not be verified by prior data. The optimal experiment corresponds to the maximum number of uncertain interactions that could be verified by the resulting data. For this purpose, we introduced the concept of edge separatoid which gave a list of nodes (genes) that upon their removal would allow the verification of a particular gene interaction. Finally, we proposed a procedure that iterates over performing KO experiments, ensemble update and optimal DOE. The case studies including the inference of Escherichia coli GRN and DREAM 4?100-gene GRNs, demonstrated the efficacy of the iterative GRN inference. In comparison to systematic KOs, REDUCE could provide much higher information return per gene KO experiment and consequently more accurate GRN estimates.ConclusionsREDUCE represents an enabling tool for tackling the underdetermined GRN inference. Along with advances in gene deletion and automation technology, the iterative procedure brings an efficient and fully automated GRN inference closer to reality.Availability and implementationMATLAB and Python scripts of REDUCE are available on www.cabsel.ethz.ch/tools/REDUCE CONTACT: rudi.gunawan@chem.ethz.chSupplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC4803391 | biostudies-literature
Project description:Design of a digital infinite-impulse-response (IIR) filter is the process of synthesizing and implementing a recursive filter network so that a set of prescribed excitations results a set of desired responses. However, the error surface of IIR filters is usually non-linear and multi-modal. In order to find the global minimum indeed, an improved differential evolution (DE) is proposed for digital IIR filter design in this paper. The suggested algorithm is a kind of DE variants with a controllable probabilistic (CPDE) population size. It considers the convergence speed and the computational cost simultaneously by nonperiodic partial increasing or declining individuals according to fitness diversities. In addition, we discuss as well some important aspects for IIR filter design, such as the cost function value, the influence of (noise) perturbations, the convergence rate and successful percentage, the parameter measurement, etc. As to the simulation result, it shows that the presented algorithm is viable and comparable. Compared with six existing State-of-the-Art algorithms-based digital IIR filter design methods obtained by numerical experiments, CPDE is relatively more promising and competitive.
| S-EPMC3394744 | biostudies-literature
Project description:BackgroundHigh-dimensional biomolecular profiling of genetically different individuals in one or more environmental conditions is an increasingly popular strategy for exploring the functioning of complex biological systems. The optimal design of such genetical genomics experiments in a cost-efficient and effective way is not trivial.ResultsThis paper presents designGG, an R package for designing optimal genetical genomics experiments. A web implementation for designGG is available at http://gbic.biol.rug.nl/designGG. All software, including source code and documentation, is freely available.ConclusionDesignGG allows users to intelligently select and allocate individuals to experimental units and conditions such as drug treatment. The user can maximize the power and resolution of detecting genetic, environmental and interaction effects in a genome-wide or local mode by giving more weight to genome regions of special interest, such as previously detected phenotypic quantitative trait loci. This will help to achieve high power and more accurate estimates of the effects of interesting factors, and thus yield a more reliable biological interpretation of data. DesignGG is applicable to linkage analysis of experimental crosses, e.g. recombinant inbred lines, as well as to association analysis of natural populations.
| S-EPMC2706229 | biostudies-literature
Project description:3D chromatin structure plays an important role in gene regulation by connecting regulatory regions and gene promoters. The ability to detect the formation and loss of these loops in various cell types and conditions provides valuable information on the mechanisms driving these cell states and is critical for understanding how long-range gene regulation works. Hi-C is a powerful technique used to characterize three-dimensional chromatin structure; however, Hi-C can quickly become a costly and labor-intensive endeavor, and proper planning is required to determine how to best use time and resources while maintaining experimental rigor and well-powered results. To facilitate better planning and interpretation of Hi-C experiments, we have conducted a detailed evaluation of statistical power using publicly available Hi-C datasets paying particular attention to the impact of loop size on Hi-C contacts and fold change compression. In addition, we have developed Hi-C Poweraid, a publicly-hosted web application to investigate these findings (http://phanstiel-lab.med.unc.edu/poweraid/). For experiments involving well-replicated cell lines, we recommend a total sequencing depth of at least 6 billion contacts per condition, split between at least 2 replicates in order to achieve the power to detect the majority of differential loops. For experiments with higher variation, more replicates and deeper sequencing depths are required. Exact values and recommendations for specific cases can be determined through the use of Hi-C Poweraid. This tool simplifies the complexities behind calculating power for Hi-C data and will provide useful information on the amount of well-powered loops an experiment will be able to detect given a specific set of experimental parameters, such as sequencing depth, replicates, and the sizes of the loops of interest. This will allow for more efficient use of time and resources and more accurate interpretation of experimental results.
| S-EPMC10055049 | biostudies-literature
Project description:BackgroundMicroarray comparative genomic hybridization (aCGH) evaluates the distribution of genes of sequenced bacterial strains among unsequenced strains of the same or related species. As genomic sequences from multiple strains of the same species become available, multistrain microarrays are designed, containing spots for every unique gene in all sequenced strains. To perform two-color aCGH experiments with multistrain microarrays, the choice of control sample can be the genomic DNA of one strain or a mixture of all the strains used in the array design. This important problem has no universally accepted solution.ResultsWe performed a comparative study of the two control sample options with a Streptococcus pneumoniae microarray designed with three fully sequenced strains. We separately hybridized two of these strains (R6 and G54) as test samples using the third strain alone (TIGR4) or a mixture of the three strains as control. We show that for both types of control it is advantageous to analyze spots in separate sets according to their expected control channel signal (5-15% AUC increase). Following this analysis, the use of a mix control leads to higher accuracies (5% increase). This enhanced performance is due to gains in sensitivity (21% increase, p = 0.001) that compensate minor losses in specificity (5% decrease, p = 0.014).ConclusionThe use of a single strain control increases the error rate in genes that are part of the accessory genome, where more variation across unsequenced strains is expected, further justifying the use of the mix control.
| S-EPMC2410139 | biostudies-literature
Project description:Primer and probe sequence designs are among the most critical input factors in real-time polymerase chain reaction (PCR) assay optimization. In this study, we present the use of statistical design of experiments (DOE) approach as a general guideline for probe optimization and more specifically focus on design optimization of label-free hydrolysis probes that are designated as mediator probes (MPs), which are used in reverse transcription MP PCR (RT-MP PCR). The effect of three input factors on assay performance was investigated: distance between primer and mediator probe cleavage site; dimer stability of MP and target sequence (influenza B virus); and dimer stability of the mediator and universal reporter (UR). The results indicated that the latter dimer stability had the greatest influence on assay performance, with RT-MP PCR efficiency increased by up to 10% with changes to this input factor. With an optimal design configuration, a detection limit of 3-14 target copies/10 μl reaction could be achieved. This improved detection limit was confirmed for another UR design and for a second target sequence, human metapneumovirus, with 7-11 copies/10 μl reaction detected in an optimum case. The DOE approach for improving oligonucleotide designs for real-time PCR not only produces excellent results but may also reduce the number of experiments that need to be performed, thus reducing costs and experimental times.
| S-EPMC4827641 | biostudies-literature
Project description:Cytosine methylation is an important DNA epigenetic modification. In vertebrates, methylation occurs at CpG sites, which are dinucleotides where a cytosine is immediately followed by a guanine in the DNA sequence from 5' to 3'. When located in the promoter region of a gene, DNA methylation is often associated with transcriptional silencing of the gene. Aberrant DNA methylation is associated with the development of various diseases such as cancer. Bisulfite sequencing (BS-seq) is the current "gold-standard" technology for high-resolution profiling of DNA methylation. Reduced representation bisulfite sequencing (RRBS) is an efficient form of BS-seq that targets CpG-rich DNA regions in order to save sequencing costs. A typical bioinformatics aim is to identify CpGs that are differentially methylated (DM) between experimental conditions. This workflow demonstrates that differential methylation analysis of RRBS data can be conducted using software and methodology originally developed for RNA-seq data. The RNA-seq pipeline is adapted to methylation by adding extra columns to the design matrix to account for read coverage at each CpG, after which the RRBS and RNA-seq pipelines are almost identical. This approach is statistically natural and gives analysts access to a rich collection of analysis tools including generalized linear models, gene set testing and pathway analysis. The article presents a complete start to finish case study analysis of RRBS profiles of different cell populations from the mouse mammary gland using the Bioconductor package edgeR. We show that lineage-committed cells are typically hyper-methylated compared to progenitor cells and this is true on all the autosomes but not the sex chromosomes. We demonstrate a strong negative correlation between methylation of promoter regions and gene expression as measured by RNA-seq for the same cell types, showing that methylation is a regulatory mechanism involved in epithelial linear commitment.
| S-EPMC5747346 | biostudies-literature
Project description:We mathematically design sets of static light scattering experiments to provide for model-independent measurements of ternary liquid mixing free energies to a desired level of accuracy. A parabolic partial differential equation (PDE), linearized from the full nonlinear PDE [D. Ross, G. Thurston, and C. Lutzer, J. Chem. Phys. 129, 064106 (2008)], describes how data noise affects the free energies to be inferred. The linearized PDE creates a net of spacelike characteristic curves and orthogonal, timelike curves in the composition triangle, and this net governs diffusion of information coming from light scattering measurements to the free energy. Free energy perturbations induced by a light scattering perturbation diffuse along the characteristic curves and towards their concave sides, with a diffusivity that is proportional to the local characteristic curvature radius. Consequently, static light scattering can determine mixing free energies in regions with convex characteristic curve boundaries, given suitable boundary data. The dielectric coefficient is a Lyapunov function for the dynamical system whose trajectories are PDE characteristics. Information diffusion is heterogeneous and system-dependent in the composition triangle, since the characteristics depend on molecular interactions and are tangent to liquid-liquid phase separation coexistence loci at critical points. We find scaling relations that link free energy accuracy, total measurement time, the number of samples, and the interpolation method, and identify the key quantitative tradeoffs between devoting time to measuring more samples, or fewer samples more accurately. For each total measurement time there are optimal sample numbers beyond which more will not improve free energy accuracy. We estimate the degree to which many-point interpolation and optimized measurement concentrations can improve accuracy and save time. For a modest light scattering setup, a sample calculation shows that less than two minutes of measurement time is, in principle, sufficient to determine the dimensionless mixing free energy of a non-associating ternary mixture to within an integrated error norm of 0.003. These findings establish a quantitative framework for designing light scattering experiments to determine the Gibbs free energy of ternary liquid mixtures.
| S-EPMC3411594 | biostudies-other