02/05/2014 /2013 MBARI Jarred Swalwell. Jarred was MBARI seminar speaker. Had discussions with Jarred at 9:00, 12:00, 16:00
 
9:00 Klimov, Swalwell, O'Reilly
 
Status: new electronics is done but have a noise on analog board from other (switching) board in a board stack. Due to limited resources, now working on components that can be used for existing systems SeaFlow (synergy). Need to solve:
Flowcell
Modified concept.
Laser (test, select)
Pump
Mechanics.
 
"Unattended" Seaflow operation of a few days is pretty reliable. Generally shorter deployments (few days, a week) compare to CytoBot
 
Rosette profiling pump may be a bit fast for cytometry. Need at least 3 minutes sampling same volume to get good population statistics
 
May come out again in April
 
Flow cell - hoping to use cheap $10 cell, as opposed to IFCB cell which costs $1000s
But have to frequently change the cell. Jarred uses MicroPump devices, pretty reliable.
Keep clean: consider hydrodynamics near flow cell walls.
 
Testing new cheap 0.5Watt  $500 Chinese-made laser - not too stable but can resolve prochlorococus. (ultrabright.com?) They also make $10k laser which is pretty stable so this is matter of price not technology.
 
Could do embedded system with Arduino
 
Camera - would like to take pictures of larger particles.
 
 
Round table discussion
 
16:00-17:00 Klimov, O'Reilly, Maughan, Swalwell, Hamilton, Chavez
 
Some of Francois' algorithms can be run in real-time
 
FC: What about machine-learning approaches? E.g. make inferences from clustering and local environment.
 
JS: Often have to specify centroids to Ribalet's algorithms at the beginning.
A lot of the clustering algorithms originated from medical applications (Fred Hutchinson Inst.). "Flowcast". Written in 'R', which is free (unlike MathLab)
 
JS: Has done some acoustic applications. Drug-delivery via bubbles, infer bubble material properties from acoustic scatter
 
JS: Note that size of virtual core can be adjusted. Larger VC results in poorer
 
JS: Have processed 2012 data - will provide to Francisco/Jason
 
Fluorescent beads tend to sink; so delivery rate is not constant, but this is not so important as long as some of them appear on cytogram for gating calibration
 
On "real time": 3 min is sort of enough for cytogram, makes 1 file. Real flow rate is 15 mL/min, virtual core flow rate about 15-30 uL / min
 
Plans: Fransico is asking what is good for Jarred: 1st, or 3-4 weeks of April?
Jarred: 1st week looks better.