ASICS: Automatic Statistical Identification in Complex Spectra

In ASICS: Automatic Statistical Identification in Complex Spectra

Description

Quantification of 1D 1H NMR spectra with ASICS method using a library of pure metabolite spectra. The method is presented in Tardivel et al. (2017).

Usage

ASICS(
  spectra_obj,
  exclusion.areas = matrix(c(4.5, 5.1), ncol = 2),
  max.shift = 0.02,
  pure.library = NULL,
  noise.thres = 0.02,
  joint.align = TRUE,
  threshold.noise = NULL,
  combine = NULL,
  add.noise = 0.15,
  mult.noise = 0.172,
  quantif.method = c("FWER", "Lasso", "both"),
  clean.thres = 1,
  ref.spectrum = NULL,
  seed = 1234,
  ncores = 1,
  verbose = TRUE
)

Arguments

spectra_obj	An object of class Spectra obtained with the function createSpectra.
exclusion.areas	Definition domain of spectra that has to be excluded for the quantification (ppm). By default, the water region is excluded (4.5-5.1 ppm).
max.shift	Maximum chemical shift allowed (in ppm). Default to 0.02.
pure.library	An object of class PureLibrary containing the reference spectra (pure metabolite spectra). If NULL, the library included in the package (that contains 191 reference spectra) is used.
noise.thres	Threshold for signal noise. Default to 0.02.
joint.align	Logical. If TRUE, information from all spectra is taken into account to align individual library.
threshold.noise	DEPRECATED, use noise.thres instead.
combine	DEPRECATED, use joint.align instead.
add.noise, mult.noise	additive and multiplicative noises. To set these noises, you can compute the standard deviation in a noisy area for add.noise or the standard deviation in a peak area for mult.noise when several spectra of the same sample are available. By default, add.noise = 0.15 and mult.noise = 0.172
quantif.method	either "FWER" to perform an independent quantification (the method available in ASICS since the beginning), "Lasso" to perform a joint quantification (all the spectra together) or "both" to perform a joint quantification after the FWER selection of the independent quantification. More details can be founded in the user's guide.
clean.thres	if quantif.method == "both" the percentage of spectra in which the metabolite needs to be identified by the FWER selection. Default to 1, i.e. metabolite is quantified if it was identified in at least 1% of the spectra.
ref.spectrum	index of the reference spectrum used for the alignment. Default to NULL, i.e. the reference spectrum is automatically detected.
seed	Random seed to control randomness in the algorithm (used in the estimation of the significativity of a given metabolite concentration).
ncores	Number of cores used in parallel evaluation. Default to 1.
verbose	A Boolean value to allow print out process information.

Value

An object of type ASICSResults containing the quantification results.

Note

Since version 2.3.1 small changes were applied in order to improve the speed of metabolite selection algorithm, which can slightly impact outputs of the method.

References

Tardivel P., Canlet C., Lefort G., Tremblay-Franco M., Debrauwer L., Concordet D., Servien R. (2017). ASICS: an automatic method for identification and quantification of metabolites in complex 1D 1H NMR spectra. Metabolomics, 13(10): 109. https://doi.org/10.1007/s11306-017-1244-5

See Also

ASICSResults pure_library createSpectra

Examples

# Import data and create object
current_path <- system.file("extdata", package = "ASICS")
spectra_data <- importSpectra(name.dir = current_path,
                     name.file = "spectra_example.txt", type.import = "txt")
spectra_obj <- createSpectra(spectra_data)

# Estimation of relative quantifications
to_exclude <- matrix(c(4.5, 10), ncol = 2)
resASICS <- ASICS(spectra_obj, exclusion.areas = to_exclude,
                  joint.align = FALSE, quantif.method = "FWER")

ASICS documentation built on Nov. 8, 2020, 8:19 p.m.