Dataset Information

Modeling Persistent Trends in Distributions.

ABSTRACT: We present a nonparametric framework to model a short sequence of probability distributions that vary both due to underlying effects of sequential progression and confounding noise. To distinguish between these two types of variation and estimate the sequential-progression effects, our approach leverages an assumption that these effects follow a persistent trend. This work is motivated by the recent rise of single-cell RNA-sequencing experiments over a brief time course, which aim to identify genes relevant to the progression of a particular biological process across diverse cell populations. While classical statistical tools focus on scalar-response regression or order-agnostic differences between distributions, it is desirable in this setting to consider both the full distributions as well as the structure imposed by their ordering. We introduce a new regression model for ordinal covariates where responses are univariate distributions and the underlying relationship reflects consistent changes in the distributions over increasing levels of the covariate. This concept is formalized as a trend in distributions, which we define as an evolution that is linear under the Wasserstein metric. Implemented via a fast alternating projections algorithm, our method exhibits numerous strengths in simulations and analyses of single-cell gene expression data.

SUBMITTER: Mueller J

PROVIDER: S-EPMC6428438 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Modeling Persistent Trends in Distributions.

Mueller Jonas J Jaakkola Tommi T Gifford David D

Journal of the American Statistical Association 20180612 523

We present a nonparametric framework to model a short sequence of probability distributions that vary both due to underlying effects of sequential progression and confounding noise. To distinguish between these two types of variation and estimate the sequential-progression effects, our approach leverages an assumption that these effects follow a persistent trend. This work is motivated by the recent rise of single-cell RNA-sequencing experiments over a brief time course, which aim to identify ge ...[more]

PMID: 30906084

Similar Datasets

Project description:Bone is one of the most common sites for cancer metastasis. Bone tissue is composed by different kinds of cells that coexist in a coordinated balance. Due to the complexity of bone, it is impossible to capture the intricate interactions between cells under either physiological or pathological conditions. Hence, a variety of in vivo and in vitro approaches have been developed. Various models of tumor-bone diseases are routinely used to provide valuable information on the relationship between metastatic cancer cells and the bone tissue. Ideally, when modeling the metastasis of human cancers to bone, models would replicate the intra-tumor heterogeneity, as well as the genetic and phenotypic changes that occur with human cancers; such models would be scalable and reproducible to allow high-throughput investigation. Despite the continuous progress, there is still a lack of solid, amenable, and affordable models that are able to fully recapitulate the biological processes happening in vivo, permitting a correct interpretation of results. In the last decades, researchers have demonstrated that three-dimensional (3D) methods could be an innovative approach that lies between bi-dimensional (2D) models and animal models. Scientific evidence supports that the tumor microenvironment can be better reproduced in a 3D system than a 2D cell culture, and the 3D systems can be scaled up for drug screening in the same way as the 2D systems thanks to the current technologies developed. However, 3D models cannot completely recapitulate the inter- and intra-tumor heterogeneity found in patients. In contrast, ex vivo cultures of fragments of bone preserve key cell-cell and cell-matrix interactions and allow the study of bone cells in their natural 3D environment. Moreover, ex vivo bone organ cultures could be a better model to resemble the human pathogenic metastasis condition and useful tools to predict in vivo response to therapies. The aim of our review is to provide an overview of the current trends in bone metastasis modeling. By showing the existing in vitro and ex vivo systems, we aspire to contribute to broaden the knowledge on bone metastasis models and make these tools more appealing for further translational studies.

Project description:Controlling for imperfect detection is important for developing species distribution models (SDMs). Occupancy-detection models based on the time needed to detect a species can be used to address this problem, but this is hindered when times to detection are not known precisely. Here, we extend the time-to-detection model to deal with detections recorded in time intervals and illustrate the method using a case study on stream fish distribution modeling. We collected electrofishing samples of six fish species across a Mediterranean watershed in Northeast Portugal. Based on a Bayesian hierarchical framework, we modeled the probability of water presence in stream channels, and the probability of species occupancy conditional on water presence, in relation to environmental and spatial variables. We also modeled time-to-first detection conditional on occupancy in relation to local factors, using modified interval-censored exponential survival models. Posterior distributions of occupancy probabilities derived from the models were used to produce species distribution maps. Simulations indicated that the modified time-to-detection model provided unbiased parameter estimates despite interval-censoring. There was a tendency for spatial variation in detection rates to be primarily influenced by depth and, to a lesser extent, stream width. Species occupancies were consistently affected by stream order, elevation, and annual precipitation. Bayesian P-values and AUCs indicated that all models had adequate fit and high discrimination ability, respectively. Mapping of predicted occupancy probabilities showed widespread distribution by most species, but uncertainty was generally higher in tributaries and upper reaches. The interval-censored time-to-detection model provides a practical solution to model occupancy-detection when detections are recorded in time intervals. This modeling framework is useful for developing SDMs while controlling for variation in detection rates, as it uses simple data that can be readily collected by field ecologists.

Project description:BackgroundAn important aspect of proteomic mass spectrometry involves quantifying and interpreting the isotope distributions arising from mixtures of macromolecules with different isotope labeling patterns. These patterns can be quite complex, in particular with in vivo metabolic labeling experiments producing fractional atomic labeling or fractional residue labeling of peptides or other macromolecules. In general, it can be difficult to distinguish the contributions of species with different labeling patterns to an experimental spectrum and difficult to calculate a theoretical isotope distribution to fit such data. There is a need for interactive and user-friendly software that can calculate and fit the entire isotope distribution of a complex mixture while comparing these calculations with experimental data and extracting the contributions from the differently labeled species.ResultsEnvelope has been developed to be user-friendly while still being as flexible and powerful as possible. Envelope can simultaneously calculate the isotope distributions for any number of different labeling patterns for a given peptide or oligonucleotide, while automatically summing these into a single overall isotope distribution. Envelope can handle fractional or complete atom or residue-based labeling, and the contribution from each different user-defined labeling pattern is clearly illustrated in the interactive display and is individually adjustable. At present, Envelope supports labeling with 2H, 13C, and 15N, and supports adjustments for baseline correction, an instrument accuracy offset in the m/z domain, and peak width. Furthermore, Envelope can display experimental data superimposed on calculated isotope distributions, and calculate a least-squares goodness of fit between the two. All of this information is displayed on the screen in a single graphical user interface. Envelope supports high-quality output of experimental and calculated distributions in PNG or PDF format. Beyond simply comparing calculated distributions to experimental data, Envelope is useful for planning or designing metabolic labeling experiments, by visualizing hypothetical isotope distributions in order to evaluate the feasibility of a labeling strategy. Envelope is also useful as a teaching tool, with its real-time display capabilities providing a straightforward way to illustrate the key variable factors that contribute to an observed isotope distribution.ConclusionEnvelope is a powerful tool for the interactive calculation and visualization of complex isotope distributions for comparison to experimental data. It is available under the GNU General Public License from http://williamson.scripps.edu/envelope/.

Dataset Information

Modeling Persistent Trends in Distributions.

Publications

Modeling Persistent Trends in Distributions.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets