ghar1821 / Chronoclust

A clustering algorithm that will perform clustering on each of a time-series of discrete datasets, and explicitly track the evolution of clusters over time.
GNU General Public License v3.0
15 stars 3 forks source link

Number of time points required #4

Open Jeff87075 opened 3 years ago

Jeff87075 commented 3 years ago

Hi, I am planning on performing temporal analysis on flow cytometry data but I only have two time states (pre and post therapy), so I was wondering would two time points be enough for running the Chronoclust workflow? On that note, I've also noticed the TrackSOM package on github and wanted to ask what is the difference between Chronoclust and TrackSOM?

ghar1821 commented 3 years ago

Heya,

2 time-points will be sufficient for both TrackSOM and ChronoClust.

TrackSOM is the updated better version of ChronoClust. It uses better and faster clustering algorithm (FlowSOM) to do the clustering phase. TrackSOM is also not as sensitive to hyperparameter settings as ChronoClust. I strongly recommend using TrackSOM than ChronoClust.