impulseFit: Fit gene expression data with an impulse model

In farrellja/URD: URD

Description

This fits either a linear, single sigmoid, or double sigmoid ("impulse") model to given x and y coordinates (e.g. expression data). This function is primarily used by geneCascadeProcess.

Usage

impulseFit(x, y, k = 50, interpolate = NULL, pulse.only = T, min.slope = 0.1)

Arguments

x	(Numeric) Expression data to fit (e.g. pseudotime)
y	(Numeric) Expression data to fit (e.g. expression level)
k	(Numeric) Number of sets of starting conditions to try
interpolate	(Numeric or NULL) If low number of data points, can interpolate them linearly to this number of points for choosing potential starting conditions. Default (NULL) does not do any interpolation. Interpolation is NOT used during actual function fitting.
pulse.only	(Logical) If TRUE, filters out double sigmoid functions that are monotonous and prefers either single sigmoid or linear fit for them instead.
min.slope	(Numeric) For pulse.only filtering, first derivative of double sigmoid function must include positive and negative values with absolute value > min.slope.

Details

Thanks to Yiqun Wang for considerable improvements to the impulse fitting functions.

Value

(Named List, length 4) Parameters & information about best fitting model:

• type: 0/1/2 = linear/single sigmoid/double sigmoid

• time.on: Gene onset pseudotimes

• time.off: Gene offset pseudotimes

• model: Parameters of the best model

References

Chechik G, Oh E, Rando O, Weissman J, Regev A, Koller D. "Activity motifs reveal principles of timing in transcriptional control of the yeast metabolic network." Nat Biotechnol. 2008 Nov;26(11):1251-9. doi: 10.1038/nbt.1499

farrellja/URD documentation built on June 17, 2020, 4:48 a.m.