I have a very generalized question. I am experiencing an issue with FlowSOM analysis which I would like to ask about. I get different results when I am running FlowSOM with/without unrelated markers (e.g. time, event length, 140 Ce etc.). IN my panel of markers there are some markers like the ones mentioned before which should not have an impact on the final outcome of FlowSOM but apparently I am getting different results when I am including these unrelated markers in the analysis along with my functional and lineage markers, compared to when I am not using these unrelated markers ```
Please let me know if anyone else experienced this issue and an explanation if known
Could you perhaps comment one why you would want to include non-biological "unrelated" channels such as time and event length for clustering in the first place?
I have never done / investigated this, however, wouldn't be surprise if the time channel and friends affect clustering. For example, fluctuations in flow rate, clotting etc. will introduce technical variation in measurements. That's why one typically i) excludes portions of the acquisition based on, e.g., time vs. channel intensity; and, ii) normalizes for time-drift based on bead standards. Related, Ce140 is a standard bead channel used for normalization, but shouldn't capture biological signal (i.e., actual protein).
To summarize: I don't think these should be used for clustering in the first place. My best guess is that their inclusion will introduce technical noise and artefacts that might (or, probably) affect clustering in an undesirable (non-biological) way.