Module : Lineage with Monocle - ComputationalSystemsBiology/Single-cell-RNA-seq GitHub Wiki

Lineage with Monocle

This module infers lineage using Monocle 2.

  • Internal name : lineage-monocle

  • Avalaible : local mode

  • Input Ports :

    • matrix : filtered expression matrix (tsv)
    • cells : normalized cells metadata (tsv)

  • Output Ports :

    • none

  • Optional parameters :

Parameter Type Description Default Value
exp_family text Expression model family, should be one of : negbinomial, negbinomial.size negbinomial
detection_threshold float Gene detection parameter 0.1
bypass text Use Monocle's native (native) normalization or bypass (bypass) using Eoulsan normalization bypass
select_genes boolean Whether to reduce gene set or not for dimensionality reduction False
mean_threshold float Minimum mean expression value to keep a gene 0.5
dispersion_fold_threshold float Minimum fold of dispersion to keep a gene 1
max_dim int Maximum number of dimensions to keep after dimensionality reduction 2
reduction_method int Dimensionality reduction method to use, should be one of : DDRTree or ICA DDRTree
norm_method string Variance stabilizing method to use, should be one of : vstExprs, log, or none vstExprs
reverse boolean Whether to represent cells in inverted order or not False
color_by string Column to use to color cells on plot State
  • Configuration example
<step id="Lineage" skip="false">
	<module>lineage-monocle</module>
	<parameters>
		<parameter>
			<name>bypass</name>
			<value>bypass</value>	
		</parameter>
		<parameter>
			<name>select_genes</name>
			<value>False</value>	
		</parameter>
		<parameter>
			<name>color_by</name>
			<value>fluorescence</value>	
		</parameter>
	  </parameters>
</step>

Interpreting output files

Mean-dispersion plot

Gene selection is conducted on log transformed data. Maximum Likelihood Estimate for mean and dispersion parameter of a Negative Binomial are calculated. The results are shown in the following plot. Only genes showing sufficient (user defined) gene expression and dispersion are kept for following analysis. Those genes appear in black. The red line is the estimate of dispersion as a function of mean.

meanVar

Lineage plot

The end product of the module is a lineage plot. Briefly, Monocle learns a minimal spanning tree from the data and then projects the tree on a two dimensional space. This tree is supposed to reflect the evolution of genes expression across an activation or a differentiation path. If the extremities of the plot are stable, branching points depend greatly on the data. We suggest to use it more as a way to challenge a hypothesis ("supposing this really occurs, does my data really reflect this process ?") than as a reliable reconstruction of the process.

lineage

⚠️ **GitHub.com Fallback** ⚠️