From: Jan Keij (jkeij@server.nybc.org)
Date: Tue Apr 11 1995 - 09:51:06 EST
Hi fellow sorters,
I have followed the exchange of data on cell rate versus detection rate for
a couple of days now and I feel compelled to suggest that the 'mailers'
read the article 'Coincidence in high-speed flow cytometry: models and
measurements' in Cytometry 12:398-404, 1991.
Yes, I realize it's my own work and it was done with a non commercial
instrument, but I believe it's relevant here.
The article describes several models that allow the analysis and prediction
of coincidence in the instrument 'dead time' (or 'cycle time' or
'interrogation' time as BD calls it) as a function of the dead time and the
cell (true throughput rate).
Lately, I have been trying to COMPLETELY understand what is going on in a
FACStar Plus as I want to get maximum performance out of it for CD34 sorts.
I have done similar measurements with yeast cells (they don't stick to
tubing) and I have obtained similar results. It should be noted that all
numbers represent the output of counters.
ANALYSIS RATES
CELL RATE vs SYS RATE
------------------------------------------------------
cell rate (/s) sys rate (/s) efficiency (%)
245 245 100
490 508 104
980 1028 105
1960 1970 100
3920 3970 101
7840 7223 92
15680 11636 74
31360 16131 51
N.B. The interrogation time was set 20 us.
As the FACStar aborts a the processing of a trigger pulse when a 2nd event
is detected within the interrogation time, the fraction of detected single
cells (=sys rate) can be approximated from:
f(1)= P(1) / (1 - P(0) or roughly f(1)= 0.5 * PW *CR,
where PW is the pulse width (20 ms) and CR is the true cell rate (cells/s).
SORT RATES
CELL RATE vs SORT RATE
------------------------------------------------------------------
cell rate (/s) sort rate (/hr) interrogation (us) eff.(%)
490 20,220 20 100
490 19,620 40 97
490 18,240 60 90
490 21,900 80 108
3920 119,400 20 74
3920 122,400 40 75
3920 150,180 60 65
3920 90,540 80 56
15680 188,520 20 29
15680 187,620 40 29
15680 114,300 60 17
15680 70,440 80 11
N.B. The sort was NORMAL-R, drop drive was 37 kHz (PW = 81 us), 3
drops/cells, and the sort fraction was 0.017.
At the high cell rates there is a dramatic drop in performance that cannot
be easily explained. I am still in discussion with Larry Duckett from BD
about what all the aborts are in the FACStar Plus and how they are
processed. It tricky.
SORT MODE vs SORT RATE
-------------------------------------------------------------
cells ENRICH (3) NORMAL-R (3) NORMAL-C (3) NORMAL-C (1)
(/s) (/hr) (/hr) (/hr) (/hr)
245 9,720 8,880 9,540 9,120
490 19,620 18,540 18,660 18,360
980 39,240 40,380 30,240 34,020
1960 70,620 65,280 42,780 60,720
3920 139,020 123,660 51,180 101,820
7840 227,940 185,100 32,460 127,380
15680 256,860 186,240 10,140 105,120
31360 160,680 111,900 660 42,000
N.B. Interrogation time was 20 us, the sort fraction was 0.017, the drop
frequency was 37 kHz and (1) and (3) represent 1 drop/cell and 3
drops/cell, respectively.
>From these data, rough estimates be made of the sort purities when the
NORMAL-R and ENRICH mode are used.
My conclusion was to give up trying to make a slow sorter sort at
high-speed by adopt a combination of immuno-magnetic bead enrichment and a
slow (< 1000/s) final purification sort with coincidence checking on and
pulse processing activated.
For applications that do not allow pre-enrichment by surface markers, the
FLU trigger method is the way to go as it saves a lot of time that is
normally wasted on baby-sitting a long sort.
Hope this is some help.
Jan
This archive was generated by hypermail 2.1.6 : Thu Jan 01 2004 - 17:30:30 EST
