fitDynamic: Fit expression dynamic

In elldc/CellTrails: Reconstruction, visualization and analysis of branching trajectories

Description

Fits feature expression as a function of pseudotime along a defined trail.

Usage

fitDynamic(sce, feature_name, trail_name)

Arguments

sce	A SingleCellExperiment object
feature_name	Name of feature
trail_name	Name of trail

Details

A trail is an induced subgraph of the trajectory graph. A trajectory graph is composed of samples (nodes) that are connected (by weighted edges) if they are chronologically related. A trail has to be defined by the user using addTrail. A pseudotime vector is extracted by computing the geodesic distance for each sample from the trail's start node. To infer the expression level of a feature as a function of pseudotime, CellTrails used generalized additive models with a single smoothing term with four basis dimensions. Here, for each feature CellTrails introduces prior weights for each observation to lower the confounding effect of drop-outs to the maximum-likelihood-based fitting process as follows. Each non-detect of feature j in state h is weighted by the relative fraction of non-detects of feature j in state h; detected values are always assigned weight = 1.

Value

An object of type list with the following components

pseudotime

The pseudotime along the trail

expression

The fitted expression values for each value of pseudotime

gam

A object of class gamObject

Author(s)

Daniel C. Ellwanger

See Also

addTrail gamObject

Examples

# Example data
data(exSCE)

# Fit dynamic
fit <- fitDynamic(exSCE, feature_name="feature_3", trail_name="Tr1")

summary(fit)

elldc/CellTrails documentation built on May 16, 2020, 4:40 a.m.