scTrace 0.1.5

scTrace+: enhance the cell fate inference by integrating the lineage-tracing and multi-faceted transcriptomic similarity information

scTrace+

Introduction

scTrace+ is a computational method to enhance the cell fate inference by integrating the lineage-tracing and multi-faceted transcriptomic similarity information.

System Requirements

• Python version: >= 3.7

Installation

The Release version scTrace+ python package can be installed directly via pip:

pip install scTrace

Besides, we provided the develop version of scTrace+. After installing scStateDynamics and node2vec, you can run our tutorial to perform LT-scSeq data enhancement and cell fate inference steps.

pip install scStateDynamics
pip install node2vec
git clone https://github.com/czythu/scTrace.git

Quick Start of LT-scSeq data enhancement

Refer to folder: tutorial for full pipeline.

Example data: Larry-Invitro-differentiation

Below are the introduction to important functions, consisting of the main steps in scTrace+.

1. prepareCrosstimeGraph: Process input time-series dataset, output lineage adjacency matrices and transcriptome similarity matrices, both within and across timepoints.

2. prepareSideInformation: Derive low-dimensional side information matrix with node2vec and rbf kernel.

3. trainMF: Train scLTMF model to predict the missing entries in the original across-timepoint transition matrix.

4. predictMissingEntries: Load pretrained scLTMF model and calculate performance evaluation indicators.

5. prepareScdobj: Prepare scStateDynamics objects and perform clustering method.

6. visualizeLineageInfo & visualizeEnhancedLineageInfo: Visualize cluster alignment results with Sankey plot.

7. assignLineageInfo: Assign fate information at single-cell level and output a cell2cluster matrix according to lineage information.

8. enhanceFate: Enhance cell fate information based on hypothesis testing method for single-cell level fate inference.

9. runFateDE: Perform differential expression analysis between selected dynamic sub-clusters.

10. dynamicDiffAnalysis: Perform differential expression analysis between all dynamic sub-clusters (1 v.s. rest).