Awesome

VeTra

A tool for infering scRNA_seq trajectory from RNA velocity

Citation

https://doi.org/10.1093/bioinformatics/btab364

Dependency

openmpi
JPype
rpy2
princurve
sklearn
scipy
numpy
statsmodels

The script works on python. R 3.6(or higher version) and "princurve" package in R are required. TF(Transfer entroy) inference needs openmpi installation on Linux OS. If you only want to infer trajectories of datasets, it is unnessary to install "openmpi" and "JPype".

1. Run VeTenet using expression data in a csv file and RNA velocity files in txt format

Initialize an example(please make your script in the same folder of VeTra to import VeTra successfully)

import VeTra as vt
ex1 = vt.VeTra("embedding.txt", "delta_embedding.txt")

Execute trajectory inference

ex1.vetra(deltaThreshold=12, WCCsizeCutoff=5, clusternumber=3,cosine_thres=0.7, expand=2)

(1) deltaThreshold - It determines how many neighbors selected for each cell. It depends on how many cells in total. For example, deltaThreshold = 20~50 is good. for the number of cells equalling to 3000。

(2) WCCsizeCutoff - It depends on the quality of rna velocity. If there are many cells with strange velocity direction, WCCsizeCutoff should be high. Since it can filter out cells of low quality.

(3) clusternumber - The number of trajectory you expect.

(4) cosine_thres - The default of cosine_thres is 0.7, which is a threshold to pick up similar neighbors of each cell. High cosine_thres usually finds less trajectories.

(5) expand - The default of expand is 2. If you find the trajectories inferred dont cover all possbile cells, you will change expand larger.

Run TENET for all trajectories

ex1.run_tenet_tf(expression="chroman_exp_filtered.csv", thread= 15, history_len = 1, species = 'mouse', bulk_run=True)

(1) history_length - the length of history. In the benchmark data TENET provides best result when the length of history set to 1.

Make GRNs for all trajectories

ex1.makeGRN_tf(method= "links", threshold = 1000, bulk_run=True)

Output

TE_result_matrix.txt - TEij, M genes x M genes matrix representing the causal relationship from GENEi to GENEj.

TE	GENE_1	GENE_2	GENE_3	...	GENE_M
GENE_1	0	0.05	0.02	...	0.004
GENE_2	0.01	0	0.04	...	0.12
GENE_3	0.003	0.003	0	...	0.001
.
.
.
GENE_M	0.34	0.012	0.032	...	0