Showing 4 of total 4 results (show query)
cdowd
twosamples:Fast Permutation Based Two Sample Tests
Fast randomization based two sample tests. Testing the hypothesis that two samples come from the same distribution using randomization to create p-values. Included tests are: Kolmogorov-Smirnov, Kuiper, Cramer-von Mises, Anderson-Darling, Wasserstein, and DTS. The default test (two_sample) is based on the DTS test statistic, as it is the most powerful, and thus most useful to most users. The DTS test statistic builds on the Wasserstein distance by using a weighting scheme like that of Anderson-Darling. See the companion paper at <arXiv:2007.01360> or <https://codowd.com/public/DTS.pdf> for details of that test statistic, and non-standard uses of the package (parallel for big N, weighted observations, one sample tests, etc). We also include the permutation scheme to make test building simple for others.
Maintained by Connor Dowd. Last updated 2 years ago.
17 stars 6.88 score 62 scripts 8 dependents
tomroh
fitur:Fit Univariate Distributions
Wrapper for computing parameters for univariate distributions using MLE. It creates an object that stores d, p, q, r functions as well as parameters and statistics for diagnostics. Currently supports automated fitting from base and actuar packages. A manually fitting distribution fitting function is included to support directly specifying parameters for any distribution from ancillary packages.
Maintained by Thomas Roh. Last updated 3 years ago.
4 stars 5.32 score 35 scripts
bioc
scShapes:A Statistical Framework for Modeling and Identifying Differential Distributions in Single-cell RNA-sequencing Data
We present a novel statistical framework for identifying differential distributions in single-cell RNA-sequencing (scRNA-seq) data between treatment conditions by modeling gene expression read counts using generalized linear models (GLMs). We model each gene independently under each treatment condition using error distributions Poisson (P), Negative Binomial (NB), Zero-inflated Poisson (ZIP) and Zero-inflated Negative Binomial (ZINB) with log link function and model based normalization for differences in sequencing depth. Since all four distributions considered in our framework belong to the same family of distributions, we first perform a Kolmogorov-Smirnov (KS) test to select genes belonging to the family of ZINB distributions. Genes passing the KS test will be then modeled using GLMs. Model selection is done by calculating the Bayesian Information Criterion (BIC) and likelihood ratio test (LRT) statistic.
Maintained by Malindrie Dharmaratne. Last updated 5 months ago.
rnaseqsinglecellmultiplecomparisongeneexpression
8 stars 4.90 score 6 scripts
hectorrdb
Ecume:Equality of 2 (or k) Continuous Univariate and Multivariate Distributions
We implement (or re-implements in R) a variety of statistical tools. They are focused on non-parametric two-sample (or k-sample) distribution comparisons in the univariate or multivariate case. See the vignette for more info.
Maintained by Hector Roux de Bezieux. Last updated 10 months ago.
1 stars 4.86 score 16 scripts 3 dependents