Last updated on 2021-02-06
by Katharine Mullen
Chemometrics and computational physics are concerned with the analysis
of data arising in chemistry and physics experiments, as well as the
simulation of physico-chemico systems. Many of the functions in base
R are useful for these ends.
The second edition of
Chemometrics with R: Multivariate Data Analysis in the Natural and Life Sciences by Ron Wehrens,
ISBN 978-3-662-62027-4, Springer, 2020, provides an introduction to
multivariate statistics in the life sciences, as well as coverage of several
specific topics from the area of chemometrics. The associated package rwehrens/ChemometricsWithR facilitates reproduction of the examples in the book.
The book
Modern Statistical Methods for Astronomy With R Applications by Eric D. Feigelson and G. Jogesh Babu, ISBN-13: 9780521767279, Cambridge, 2012,
provides an introduction to statistics for astronomers and an
overview of the foremost methods being used in astrostatistical analysis,
illustrated by examples in R.
The book by Kurt Varmuza and Peter Filzmoser,
Introduction to Multivariate Statistical Analysis in Chemometrics,
ISBN 978-1-420-05947-2, CRC Press, 2009, is associated with the
package chemometrics.
A special issue of R News with a focus on
R in Chemistry
was published in August 2006. A special volume of Journal of Statistical Software (JSS) dedicated to
oscopy and Chemometrics in R
was published in January 2007.
Please let us knowif
we have omitted something of importance, or if a new package or function
should be mentioned here.
Linear Regression Models
- Linear models can be fitted (via OLS) with
lm()
(from stats). A least squares solution for x in
Ax = b can also be computed as
qr.coef(qr(A), b).
- The package nnls provides a means of constraining
x
to non-negative or non-positive values; the package
bvls allows other bounds on x to be applied.
- Functions for isotonic regression are available in the package Iso,
and are useful to determine the unimodal vector that is closest to
a given vector
x under least squares criteria.
- Heteroskedastic linear models can be fit using the
gls() function of the nlme package.
Nonlinear Regression Models
- The
nls() function
(from stats) as well as the package
minpack.lm allow the solution of nonlinear
least squares problems.
- Correlated and/or
unequal variances can be modeled using the
gnls() function of the nlme package
and by nlreg.
Curve Resolution
- The PTAk package provides functions for
Principal Tensor Analysis on k modes.
The package includes also some other multiway methods:
PCAn (Tucker-n) and PARAFAC/CANDECOMP.
- Multivariate curve resolution alternating least squares (MCR-ALS)
is implemented in the package ALS.
- The alsace package provides MCR-ALS support for Liquid chromatography with PhotoDiode Array Detection
(LC-DAD) data with
many injections, with features for peak alignment and identification.
- The package drc provides functions for the analysis
of one or multiple non-linear curves with focus on models for
concentration-response, dose-response and time-response data.
Partial Least Squares
- The package pls implements
Partial Least Squares Regression (PLSR) and Principal
Component Regression (PCR).
- The package
lspls implements the
least squares-partial least squares (LS-PLS) method.
- Sparse PLS is implemented in the package
spls package.
- The gpls package implements
generalized partial least squares, based on the Iteratively
ReWeighted Least Squares (IRWLS) method of Brian Marx.
- The package enpls implements ensemble partial
least squares, a
framework for measuring feature importance, outlier detection,
and ensemble modeling based on (sparse) partial least squares
regressions.
Principal Component Analysis
- Principal component analysis (PCA) is in the package stats as functions
princomp(). Some graphical PCA representations can be
found in the psy package.
- The homals package provides nonlinear
PCA and, by defining sets, nonlinear canonical
correlation analysis (models of the Gifi-family).
- A desired number of robust principal components can be computed
with the pcaPP package. The package elasticnet
is applicable to sparse PCA. The package fpca
can be applied to restricted MLE for functional PCA.
- The subselect provides a collection of functions
which assess the quality of variable subsets as surrogates for a full
data set.
- See the Multivariate task view for further packages dealing with
PCA and other projection methods.
Factor Analysis
- Factor analysis (FA) is in the package stats as functions
factanal(); see Psychometrics
task view for details on extensions.
Compositional Data Analysis
- The package compositions
provides functions for the consistent analysis of compositional data (e.g. portions of substances) and positive numbers (e.g. concentrations).
See also the book, Analyzing Compositional Data with
R by K. Gerald von den Boogaart und Raimon Tolosana-Delgado,
ISBN: 978-3-642-36808-0, Springer, 2013.
Independent Component Analysis
- Independent component analysis (ICA) can be computed using
fastICA.
Clustering
- The Cluster task view provides a list of packages that can be
used for clustering problems.
Variable Selection
- Stepwise variable selection for linear models, using AIC, is available
in function
step(); package leaps implements
leaps-and-bounds variable
selection, by default using Mallow's Cp. stepPlr provides
stepwise variable selection for penalized logistic regression.
- Package
varSelRF provides variable selection methods for random
forests. Cross-validation-based variable selection using Wilcoxon rank
sum tests is available in package WilcoxCV, focused on
binary classification in microarrays. Package clustvarsel
implements variable selection for model-based clustering.
-
The BioMark package
implements two meta-methods for variable selection: stability selection (applying a primary selection method like a t-test, VIP value or PLSDA regression coefficient) to different subsets of the data, and higher criticism, which provides a data-driven choice of significance cutoffs in statistical testing.
Self-Organizing Maps
- The kohonen package implements self-organizing maps as well as
some extensions for supervised pattern recognition and data fusion.
The som package provides functions for self-organizing maps.
Differential Equations
Metrology
- The units package attaches unit metadata to vectors, matrices and arrays,
providing automatic propagation, conversion, derivation and
simplification of units.
- The errors attaches uncertainty metadata to vectors, matrices and
arrays, providing automatic propagation and reporting.
- The constants package provides values of the fundamental physical constants based on values reported by the Committee on Data for Science and Technology (CODATA),
an interdisciplinary committee of the International Council for Science.
- NISTunits also provides values of the fundamental physical constants. The values it contains are based on the values reported by the National Institute of Standards and Technology, (NIST).
- The measurements contains tools to make working with physical measurements
easier, such as functions to convert between metric and imperial units, or to calculate a dimension's
unknown value from other dimensions' measurements.
- The metRology package provides support
for metrology applications, including measurement uncertainty estimation
and inter-laboratory metrology comparison studies.
- The ATmet package provides functions for smart sampling and sensitivity analysis for metrology applications, including computationally expensive problems.
Calibration
- The investr package facilitates calibration/inverse
estimation with linear and nonlinear regression models.
- The chemCal package provides functions for plotting
linear calibration functions and estimating standard errors for
measurements.
-
The
nlreg package is useful for nonlinear calibration models.
- The package represent calculates the 'representativity'
of two multidimensional
datasets, which involves comparison of the similarity of principal component
analysis loading patterns, variance-covariance matrix structures,
and data set centroid locations.
Cellular Automata
- The simecol package includes functions for cellular automata
modeling.
Thermodynamics
- The CHNOSZ package provides functions
for calculating the standard Gibbs energies and
other thermodynamic properties, and chemical affinities of reactions
between species contained in a thermodynamic database.
Interfaces to External Libraries
- The package rcdk allows
the user to access functionality in the
Chemistry Development Kit (CDK),
a Java framework for cheminformatics. This allows the
user to load molecules, evaluate fingerprints (via the package
fingerprint), calculate molecular
descriptors and so on. In addition, the CDK API allows the user to
view structures in 2D. The rcdklibs package provides the CDK
libraries for use in R.
-
ChemmineR is
a cheminformatics toolkit for analyzing small molecules in R. Its add-on
packages include fmcsR for
mismatch tolerant maximum common substructure matching, eiR
for
accelerated structure similarity searching; bioassayR
for analyzing bioactivity data, and ChemmineOB
for accessing OpenBabel
functionalities from R.
-
The webchem package allows users to retrieve chemical information
from various sources on the web and to interact with various APIs. Sources
include:
Chemical Identifier Resolver,
ChemSpider,
PubChem,
Chemical Translation Service,
PAN Pesticide Database,
Alan Wood's Compendium of Pesticide Common Names,
PHYSPROP Database,
ETOX,
PPDB, and
ChemIDplus.
Spectroscopy
-
Bryan Hanson has compiled a broad range of
Free and Open Source Software (FOSS) for Spectroscopy, much of which is in the form of R packages.
-
The spectralAnalysis package allows users to pre-process, visualize and analyze spectroscopy data. Non-negative matrix factorization analysis is included.
- The ChemoSpec package collects user-friendly
functions for plotting spectra (NMR, IR, etc) and
carrying top-down exploratory data analysis, such as HCA, PCA
and model-based clustering.
- The Chathurga/HyperChemoBridge interconverts ChemoSpec (and hyperSpec) objects
- The speaq package implements the hierarchical Cluster-based Peak Alignment (CluPA) and may be used for aligning NMR spectra.
- The package TIMP
provides a problem solving environment for fitting
separable nonlinear models in physics and chemistry applications, and has been
extensively applied to time-resolved spectroscopy data.
- The package ChemoSpec2D
allows exploratory chemometrics of 2D spectroscopic data sets such as COSY (correlated spectroscopy) and HSQC (heteronuclear single quantum coherence) 2D NMR (nuclear magnetic resonance) spectra.
- The spectrino package provides tools for spectra viewing and organization.
Mass Spectrometry
- The MSnbase defines infrastructure for
mass spectrometry-based proteomics data handling,
plotting, processing and quantification.
- The MALDIquant provides tools for quantitative analysis
of MALDI-TOF mass spectrometry data, with support for
baseline correction, peak detection and plotting of mass spectra.
- The OrgMassSpecR package
is for organic/biological mass spectrometry, with a focus on
graphical display, quantification
using stable isotope dilution, and protein hydrogen/deuterium
exchange experiments.
- The Bioconductor packages
MassSpecWavelet, PROcess, and xcms
are designed for the analysis of mass spectrometry data.
- The apLCMS
package is designed for the processing of LC/MS based metabolomics data.
- The xMSanalyzer
package allows merging apLCMS
sample processing results from multiple sets of parameter
settings, among other features.
- The MSPrep
package is for post-processing of metabolomic data, including summarization of replicates, filtering, imputation, and normalization.
- The metaMS package is an MS-based metabolomics data processing and compound annotation pipeline.
Functional Magnetic Resonance Imaging
- Functions for I/O, visualization and analysis of functional
Magnetic Resonance Imaging (fMRI) datasets stored in the ANALYZE
or NIFTI format are available in the package AnalyzeFMRI.
The package fmri contains functions to analyze fMRI data using
adaptive smoothing procedures.
Fluorescence Lifetime Imaging Microscopy
- Functions for visualization and analysis of
Fluorescence Lifetime Imaging Microscopy (FLIM)
datasets are available in the package TIMP.
Fluorescence Excitation-Emission Matrix (EEM)
- The EEM reads raw EEM
data and prepares it for further analysis.
Carbon Dating
- The package Bchron creates
chronologies based on radiocarbon and non-radiocarbon dated depths.
Astronomy and astrophysics
- The astrodatR package collects 19 datasets from
contemporary astronomy research, many of which are described in the aforementioned textbook ‘Modern Statistical Methods for Astronomy with R Applications’.
- The astrolibR package presents an R interface to low-level utilities and codes from the
Interactive Data Language (IDL) Astronomy Users Library.
- The CRAC collects R functions for cosmological research, with
its main functions being similar to the python library, cosmolopy.
- The RobPer package calculates periodograms based on (robustly) fitting periodic functions to light curves.
- The package snapshot contains functions for reading and writing N-body snapshots from the GADGET code for cosmological N-body/SPH simulations.
- The package UPMASK performs unsupervised photometric membership assignment in stellar clusters using, e.g., photometry and spatial
positions.
- The solaR package provides functions to determine the movement of the sun from
the earth and to determine incident solar radiation.
- The FITSio package provides utilities to read and write files in the FITS (Flexible Image Transport System) format, a standard format in astronomy.
- The stellaR package manages and displays stellar tracks and isochrones from the Pisa low-mass database.
- The astroFns provides miscellaneous astronomy functions, utilities, and data.
- The cosmoFns contains standard expressions for
distances, times, luminosities, and other quantities useful in
observational cosmology, including molecular line observations.
- The celestial package includes a number of common astronomy conversion routines, particularly the HMS and degrees schemes.
- The SCEPtER package
is used to
estimate stellar mass and radius given observational data of effective
temperature, [Fe/H], and astroseismic parameters.
- The lira package performs Bayesian linear regression and forecasting in Astronomy, accounting for all kinds of errors and correlations in the data.
- The SPADAR package provides functions to create all-sky grid plots of widely used astronomical coordinate systems (equatorial, ecliptic, galactic) and scatter plots of data on any of these systems, including on-the-fly system conversion.
- The SCEPtERbinary allows for estimating the stellar age for double-lined detached binary systems, adopted from the effective temperature, the metallicity [Fe/H], the mass, and the radius of the two stars.
- The Astrostatistics and Astroinformatics Portal is an R-centric collection of information regarding statistical analysis in astronomy.
- Hans Werner Borchers has a page on Astronomy modules and links for R, Python, and Julia.
Optics and Scattering Approximations
- The planar package provides code to simulate
reflection and transmission at a multilayer planar interface.
- The dielectric package defines some physical constants and dielectric functions commonly used in optics and plasmonics.
Energy Modeling
- The solaR package provides functions to simulate and model systems involved in
the capture and use of solar energy, including
photovoltaics.
Water and Soil Chemistry
- The AquaEnv package is a toolbox for aquatic
chemical modelling focused on (ocean) acidification and CO2 air-water
exchange.
- See the Environmetrics task view for further related
packages related to water and soil chemistry.
Titration Curves
- The titrationCurves package provides functions to plot
acid/base, complexation, redox, and precipitation titration curves.
Electrochemistry
- The eChem package provides functions to simulate
voltammetry, chronoamperometry and chronocoulometry experiments,
which may be useful in courses in
analytical chemistry.
Health Physics
- The package radsafer
provides functions for radiation safety; the package RadData
provides
nuclear decay data for dosimetric calculations from the
International Commission on Radiological Protection.