Table 1(continued)Tool
Description
Requirements
Type
Interface
Operatingsystem
Programinglanguages References
Destiny
Extends diffusion maps
(a nonlinear dimensionreduction approach) to handlezeros and sampling densityheterogeneities inherent insingle-cell data
Requires variance-stabilized
gene expression estimates;works best with a largenumber of cells
Package
Command
line
Unix/
Linux,Mac OS,Windows
R[
99
]
SNN-Cliq
Clusters cells by identifying and
merging subgraphs (quasi-cliques) in a shared nearestneighbor (SNN) graph; thenumber of clusters is chosenautomatically
Requires a reduced set of genes.
The authors recommendusing genes with averageRPKM
>
20 and using a log
transformation to reduce theeffect of outliers. Relies on avalid choice of graphparameters
Package
Command
line
Unix/
Linux
MATLAB,
Python
[^100
]
SCUBA
Uses k-means to cluster data
along a binary tree detailingbifurcation events for timecourse data. Modelsexpression regulation alongthe tree using bifurcationtheory
Requires a reduced set of genes.
The authors recommendusing the 1000 most variablegenes that are expressed in atleast 30% of cells
Package
Command
line
Unix/
Linux,Mac OS,Windows
MATLAB [
101
]
Cell lineage and pseudotime inferenceMonocle
Reduces data using independent
component analysis (ICA) andconstructs a minimumspanning tree (MST) to ordercells in pseudotime
Requires normalized,
log-transformed geneexpression estimates and areduced set of genes. Theauthors recommend
Toolkit
Command
line
Unix/
Linux,Mac OS
[^30
]
344 Yungang Xu and Xiaobo Zhou