sqjin / scepath

An energy landscape-based approach for measuring developmental states and inferring cellular trajectories from single cell RNA-seq data

scEpath

Package of scEpath (a novel tool for analyzing single cell RNA-seq data)

Overview

This is a MATLAB Package of scEpath ("single-cell Energy path"). scEpath is a novel computational method for quantitatively measuring developmental potency and plasticity of single cells and transition probabilities between cell states, and inferring lineage relationships and pseudotemporal ordering from single-cell gene expression data. In addition, scEpath performs many downstream analyses including identification of the most important marker genes or transcription factors for given cell clusters or over pseudotime.

The rational of scEpath for inferring cellular trajectories is based on the famous Waddington's landscape metaphor for describing the cellular dynamics during the development. Below is a conceptual illustration from a paper (Takahashi et al. Development, 2015)

Check out our paper (Jin et al. Bioinformatics, 2018) for the detailed methods and applications. Below is the overview of scEpath.

Systems Requirements

scEpath is independent of operating systems because it is written in Matlab. Basic requirement for running scEpath includes MATLAB and the Statistics toolbox. The pseudotime estimation step requires the R package "princurve" for principal curve analysis. In this case, both R and Matlab are required for running scEpath.

This Package has been tested using MATLAB 2016a/b/2017a on Mac OS/64-bit Windows.

Usage

Unzip the package. Change the current directory in Matlab to the folder containing the scripts.

This directory includes the following main scripts:

1. scEpath_demo.m -- an example run of scEpath on a specific dataset
2. preprocessing.m -- do preprocessing of the input data (if applicable)
3. constructingNetwork.m -- construct a gene-gene co-expression network
4. estimatingscEnergy.m -- estimate the single cell energy (scEnergy) for each cell
5. ECA.m -- prinpipal component analysis of energy matrix
6. clusteringCells.m -- perform unsupervised clustering of single cell data
7. addClusterInfo.m -- integrate clustering information
8. inferingLineage.m -- infer the cell lineage hierarchy
9. FindMDST.m -- find the minimal directed spanning tree in a directed graph
10. inferingPseudotime.m -- reconstruct pseudotime
11. smootheningExpr.m -- calculating the smooth version of expression level based on pseudotime
12. identify_pseudotime_dependent_genes.m -- identify pseudotime dependent marker genes
13. identify_keyTF.m -- identify key transcription factors responsible for cell fate decision

---

14. cluster_visualization.m -- visualize cells on two-dimensional space
15. lineage_visualization.m -- display cell lineage hierarchy with transition probability
16. scEnergy_comparison_visualization.m -- comparison of scEnergy among different clusters
17. landscape_visualization -- display energy landscape in 2-D contour plot and 3-D surface
18. plot_genes_in_pseudotime.m -- plot the temporal dynamics of individual gene along pseudotime
19. plot_rolling_wave.m -- create "rolling wave" showing the temporal pattern of pseudotime-dependent genes and display gene clusters showing similar patterns
20. plot_rolling_wave_TF.m -- create "rolling wave" showing the temporal pattern of key transcription factors

For each run, the final results of the analysis are deposited in the "results" directory:

1. results/figures, containing PDF figures of the analysis.
2. results/PDG_in_each_cluster, containing the identified pseudotime-dependent marker genes in each cluster
3. results/temporalfiles, containing intermediate results from the analysis.

Please refer to scEpath_demo.m for instructions on how to use this code. Input Data are gene expression data matrix (rows are genes and columns are cells).

If you have any problem or question using the package please contact [email protected]

scepath's Issues

Error in FindMDST function

Whenever I try running the scePathDemo file, I get an error when running the FindMDST function.

The error is: Undefined function 'incidence_to_3n' for input arguments of type 'double'.

Any ideas why this might be happening?

asking for running SIMLR

hi,
I am sorry, I not good at matlab. because my cell number are with more than 3,000 cells, so I want to use the the function called "SIMLR_LARGE". I change the "SIMLR_LARGE" to "SIMLR_LARGE" in the clusteringCells.m, but I run the scEpath_demo.m that had an error the function "mex_top_eig_svd" can not identify. would be very grateful indeed for any help you could give me.

Typo in demo script (Line 133, comment 'mouse' -> 'human'

Line 133 of the demo script says

TF = TF_Ifo.human.Symbol; % TF for mouse

but should read

TF = TF_Ifo.human.Symbol; % TF for human

scEpath R package

Hello there,

scEpath looks like a really nice package to go hand-in-hand with a psuedotime package like Monocle 2. I came across your package while reading a recent paper:

https://www.nature.com/articles/s41467-018-08247-x

I'm really interested in trying out scEpath for my data. However, since the package is in Matlab, it seems hard to try out quickly, as I'm hesitant that it may not integrate well with the rest of my R-based single-cell pipeline.

By any chance, is this available in R? Or, do you plan to bring it out in R anytime soon?

Thanks!