search

hms-dbmi / scde

R package for analyzing single-cell RNA-seq data
http://pklab.med.harvard.edu/scde
Other
170 stars 64 forks source link

scDE is too slow when running large scale datasets. #78

Closed iyhaoo closed 5 years ago

iyhaoo commented 5 years ago

Hi there, scDE cost more than 3 hrs when input dataset has more than 3000 cells. I used 24 cores but the processing is still rather long. There is no doubt that scDE provides very accurate results. If the time consumption can be much lower this software will be perfect!

JEFworks commented 5 years ago

Indeed, when Dr. Kharchenko and I first developed SCDE, single-cell sequencing via the 96-well plate was still the most common. We knew 384-well plates were coming, but we never expected droplet-based microfluidics to become so readily available and so quickly!

The Kharchenko lab is actively developing PAGODA2 (as a successor to SCDE) to more appropriately accommodate newer single-cell datasets with 1000s of cells: https://github.com/hms-dbmi/pagoda2 While I am not personally involved in that effort, I'm sure you will be able to find more support on the PAGODA2 github.

Best, Jean