search

nolanlab / spade

SPADE: Spanning Tree Progression of Density Normalized Events
Other
46 stars 23 forks source link

Minimum Requirements for SPADE #133

Open miguel1velazquez opened 7 years ago

miguel1velazquez commented 7 years ago

This might be a basic question: What makes a population SPADEable if you will? What kind of threshold or minimum cells/attributes need to be filled?

Cheers, Miguel

zbjornson commented 7 years ago

Not sure I quite understand. Is there a particular shape of dataset that you have in mind? The same basic principles from gating generally apply to clustering: a single cell in a population (be it a gate or a cluster or a group of similar clusters) is harder to trust than many cells in a population, so you need to consider if your sample is homogeneous or heterogeneous and how many populations/clusters you expect to get, relative to how many cells you measured. Likewise, clustering doesn't typically reduce the amount of information that you need in order to identify populations; you still need to measure a marker unique to T cells in order to identify a T cell, although clustering might reveal a marker for T cells that you would not have typically used (e.g. they might segregate on the basis of size and a dozen other markers, none of which are canonical T cell markers). Does that help?