Developmental SpatioTemporal Optimal Transport
A method for aligning spatially resolved transcriptomics time-series.
There are four main functions:
-
src/destot/DESTOT/align: Given a pair of ST slices from two developmental timepoints, infer a spatiotemporal alignment matrix Pi and a growth vector xi. As discussed in the paper, there are two settings we recommend for this function: the default setting for growth-rates and alignments ($\alpha = 0.2$ ,$\beta = 0.5$ ,$\gamma = 50$ ,$\epsilon = 0.1$ ), and the robust setting ($\alpha = 0.99$ ,$\beta = 0.6$ ,$\gamma = 1$ ,$\epsilon = 0.1$ ) for spatiotemporal alignments with very different geometries (e.g. with capture-frame effects). -
src/destot/DESTOT/xi_to_growth_rate: Given a growth vector xi, convert the values in the growth vector to a per-spot growth rate$J$ given the start and end timepoints. -
src/destot/metrics/growth_distortion_metric: Given a pair of ST slices, their spatiotemporal alignment matrix Pi, and the inferred growth vector xi, calculcate the growth distortion metric as in Eq. 9 of the paper. -
src.destot/metrics/migration_metric: Given a pair of ST slices and their spatiotemporal alignment matrix Pi, calculate the migration metric as in Eq. 11 of the paper.
We will soon make DeST-OT available on PyPi. In the mean time, you can download the repository and call the functions directly.
If you encounter any problem running the software, please contact Xinhao Liu at [email protected] or Peter Halmos at [email protected]
Halmos, P., Liu, X., Gold, J., Chen, F., Ding, L., and Raphael, B. J. DeST-OT: Alignment of Spatiotemporal Transcriptomics Data. Cell Systems, January 2025. ISSN 2405-4712. doi: 10.1016/j.cels.2024.12.001. URL http://dx.doi.org/10.1016/j.cels.2024.12.001.
The paper is available here: <DeST-OT: Alignment of Spatiotemporal Transcriptomics Data.>, and a Zenodo registered DOI is available in the link below
