search

frisen-lab / TREX

Simultaneous lineage TRacking and EXpression profiling of single cells using RNA-seq
MIT License
5 stars 6 forks source link

The output of the clone calling algorithm does not match R output in some cases #7

Closed Leonievb closed 1 year ago

Leonievb commented 2 years ago

In some cases the clone calling algorithm produces unreasonably big clones clumping together many cells that are split into separate clones by the R algorithm (with same parameters)

marcelm commented 1 year ago

To summarize your intense research on this (which you sent me an e-mail about), the output of our clone calling algorithm exactly matches the R output if the Jaccard threshold is set to 0.7 as in R. Our default is 0.0 and, as you also suggest, we should possibly consider changing it to 0.7.

Leonievb commented 1 year ago

That’s the correct summary. I would change the jaccard score to 0.7 as default.