This repository contains a Matlab implementation of MAPiT for the transformation of pseudotime data to meaningful scales. Application cases in the associated manusucript "Reconstructing temporal and spatial dynamics in single-cell experiments" are demonstrated in example code.
MAPiT features include
- Transformation of pseudotime trajectories to real-time scales
- Reconstruction of spatial arrangement of tumor spheroids
- Easy integration with pseudotime algorithms
MAPiT itself is not a software package that has to be installed, but consists of a set of Matlab scripts that have to be adapted and called within a Matlab session for each application case.
MAPiT has no dependencies to other third-party Matlab interfaces/toolboxes. However, the examples use Wanderlust (part of the Cyt3 toolbox https://github.com/dpeerlab/cyt3) and Diffusion Maps (https://www.helmholtz-muenchen.de/icb/research/groups/machine-learning/projects/dpt/index.html) to derive pseudotime values from single-cell experimental data. Furthermore, kernel density estmation with linked boundary conditions (https://github.com/MColbrook/Kernel-Density-Estimation-with-Linked-BCs.git) is used to obtain distribution in pseudotime for the cell cycle example.
This repository contains the scripts necessary to perform the nonlinear transformation from pseudotime to real-time, or spatial scale with MAPiT, and associated workflows for all examples presented in the manuscript. The examples are:
- Toy example
example_toy.m - Cell cycle analysis
example_cellcycle.m - Spheroid analysis
example_spheroid.m - Paper Fig 2: Cell cycle - MAPiT vs. microscopy
MAPiT_CellCycle_Validation.m - Paper Fig 4: Spheroid - MAPiT vs. microscopy
MAPiT_Spheroid_Validation.m
Workflow for analysing single-cell data with MAPiT
- Generate pseudotemporal ordering of cells with your favorite algorithm
- Define true-scale distribution
- Get joint distribution of pseudotime and markers with
jointDensityPseudotimeY.m - Get transformation with
preMAPiT.m - Transform pseudotime trajectories to new scale with
MAPiT.m
Usage of MAPiT with R or Python based pseudotime analysis methods is straight
forward. Entry point of pseudotime values and single cell data is step 3.
The function jointDensityPseudotimeY.m takes a nx1 vector of
pseudotime values from the pseudotime algorithm and a nx1 vector of a marker signals from n single cells of the single cell
dataset as input. These vectors, must be imported to Matlab, e.g. by importing
a .csv file.
Citeable DOI for the latest MAPiT release:
When using MAPiT in your project, please cite
Reconstructing temporal and spatial dynamics from single-cell pseudotime using prior knowledge of real scale cell densities
Karsten Kuritz, Daniela Stöhr, Daniela Maichl, Nadine Pollak, Markus Rehm, Frank Allgöwer
Sci Rep 10, 3619 (2020). https://doi.org/10.1038/s41598-020-60400-z
