A graphical and automated pipeline for the analysis of short time-series in R ('santaR'). This approach is designed to accommodate asynchronous time sampling (i.e. different time points for different individuals), inter-individual variability, noisy measurements and large numbers of variables. Based on a smoothing splines functional model, 'santaR' is able to detect variables highlighting significantly different temporal trajectories between study groups. Designed initially for metabolic phenotyping, 'santaR' is also suited for other Systems Biology disciplines. Command line and graphical analysis (via a 'shiny' application) enable fast and parallel automated analysis and reporting, intuitive visualisation and comprehensive plotting options for non-specialist users.
Interactive package for Short AsyNchronous Time-series Analysis (SANTA), implemented in
Longitudinal studies in Systems Biology face multiple challenges that are not suitably addressed by current time-series statistical methods: it is difficult to simultaneously account for biological variability, measurement error, missing observations, asynchronous sampling, nonlinearity and low number of time points (e.g. 4-10) with a high number of variables.
To address these challenges,
santaR (Short AsyNchronous Time-series Analysis) provides a Functional Data Analysis (FDA) approach -where the fundamental units of analysis are curves representing each individual across time-, in a graphical and automated pipeline for robust analysis of short time-series studies.
Analytes levels are descriptive of the underlying biological state and evolve smoothly through time. For a single analyte, the time trajectory of each individual is described with a smooth curve estimated by smoothing splines. For a group of individuals, a curve representing the group mean trajectory is also calculated. These individual and group mean curves become the new observational unit for subsequent data analysis, that is, the estimation of the intra-class variability and the identification of trajectories significantly altered between groups.
Designed initially for metabolomic,
santaR is also suited for other Systems Biology disciplines. Implemented in
santaR is developed as a complete and easy-to-use statistical software package, which enables command line and GUI analysis, with fast and parallel automated analysis and reporting. Comprehensive plotting options as well as automated summaries allow clear identification of significantly altered analytes for non-specialist users.
Install the development version of the package directly from GitHub with:
If the dependency
pcaMethods is not successfully installed,
Bioconductor must be added to the default repositories with:
To get started
santaR's graphical user interface implements all the functions for short asynchronous time-series analysis:
library(santaR)santaR_start_GUI(browser = TRUE)# To exit press ESC in the command line
The graphical user interface is divided in 4 sections, corresponding to the main steps of analysis:
Import, DF search, Analysis and Export:
More information is available in the graphical user interface as well as in the following vignettes:
A dataset containing the concentrations of 22 mediators of inflammation over an episode of acute inflammation is also available. The mediators have been measured at 7 time-points on 8 subjects, concentration values have been unit-variance scaled for each variable. A subset of the data is presented below:
The GUI is to be prefered to understand the methodology, select the best parameters on a subset of the data before running the command line, or to visually explore results.
If a very high number of variables is to be processed,
santaR's command line functions are more efficient, as they can be integrated in scripts and the reporting automated.
santaR is licensed under the GPLv3
As a summary, the GPLv3 license requires attribution, inclusion of copyright and license information, disclosure of source code and changes. Derivative work must be available under the same terms.
© Arnaud Wolfer (2017)