[harrison]

Hey Mirren
Good question. We have not tried this in our lab ourselves, but I do know some other people have. The most detailed writeup I have seen is this one:
https://www.nature.com/articles/s41467-022-31033-9
Where they use a liquid handling robot. Your fraction collector seems sensible/viable; you might want to use slightly more sensitive/precise pumps than the chi.bio ones though. FOr example there are fairly cheap stepper motor based peristaltic punmps you could integrate, and then put an extra tube down through the chi.bio lid to be a little lower than the usual extraction tube so you can remove small volumes.

[harrison]

Hey Bill
On the side of the ChiBio there is a white header with I think 6 pins which you have likely looked at but never used. THis is a generic expansion header which includes pinout of I2C. So what I would suggest is go download the circuit schematics available on this site, figure out the pinout there, and connect directly to this. It gives you power too if you want it.

[harrison]

Have just emailed you!

[harrison]

Hey!
We have done this. I think the headspace is generally fine since the pump removing liquid in practice does refresh that air. Instead, what we have done for e.g. cyanobacteria or others where we want maximum oxygenation, is using a cheap amazon aquarium pump to bubble air into the reactor liquid by putting a longer silicone tube from one of the ports in the lid to the bottom of the test tube. Note that this can then mess with OD readings if it gets in the way but in our case this was fine. This approach (e.g. sparging gass into the liquid) was better in our opinion compared to just pushing air through the headspace (since this is done anyway to some extent)

[harrison]

Cool, I am not sure actually how best to have that work. You can remotely run it through an ethernet port into the device which works fine for us. SImilarly, we often remote in to the computer acting as USB host which also is stable. Not sure about the more general Cloud9 route – haven’t really tried it!

[harrison]

Hello, these issues seem consistent with a problem such as the host PC not having internet access – as the GUI uses various online JS packages to render the interface. Is this the case? (e.g. can you navigate to google.com on whatever computer is showing the GUI?)

[harrison]

Ah that is a very good question. Is this also happening if connecting via USB?
If I were to guess it might be that when it is outputing the files that normally go to the GUI, it is creating something like a memory leak e.g. the data is stacking up and not being cleared and so eventually there is so much to transmit it doesn’t achieve this within whatever the default timeout is and so crashes. How to fix, not sure. Maybe see if you can monitor memory usage on the beaglebone or that received in your browser on your interface end (can be done in most browsers) and check to see if memory usage is linearly increasing over time?

[harrison]

Hello Gaby
Do you have specific experience finding that pumps like this cannot operate in such circumstances? We have had these running in 37 degree hot rooms with high moisture content for some years without major drama. Further, if they DO break you can easily replace them, the pump parts themselves (minus all the control electronics) can be purchased on Alibaba/Aliexpress for <10 USD each. Alas I do not know of specific brushless motor pumps that could be substituted.
Harrison

[harrison]

OK, yes indeed you could just tell the device to stick at 0.5 speed and NOT turn off the stirring at any point.
Another thing is to make sure your tubes are going RIGHT to the bottom of the device and touching the heat pad (check with an empty/no water tube) as if they are getting stuck above this it may be they are too far from the rotating magnets…

[harrison]

Hello
Are you using the standard recommended stir bar? This behaviour is something we have observed if using a different (often much larger or less smooth) stir bar. Also, it will be more reliable if you run at higher than 0.5 speeds where possible…

[harrison]

What you are suggesting sounds broadly sensible.
To make the pip iinstall work you need to do the instructions in the software setup document about sharing internet connection through USB so the beaglebone can directly access internet. Have you done this? If you CAN’T ping 8.8.8.8 or ping google.com then it is certainly not working. There is various troubleshooting available for this online (i.e. it is a generic beaglebone setup issue – not to do with ChiBio software itself).

Then, yes, I would say you could either adjust the setup thing as you desire, or just manually go and install packages on the device. Unless you are going to be very often flashing the device it seems reasonable to do it by hand rather than faffing with automation…

[harrison]

FOr this, look at the Chi.BIo paper and consider using the Zigzag / dither OD mode. It is exactly for this kind of thing.

[harrison]

Perfect – and if that is your goal you don’t care AT ALL about the absolute OD number, just however long it takes to double. So it should very much be fine.

[harrison]

Hello
Yes I think the FigS10C is fine. THose parameters in both cases were found by calibrating against different spectrophotometers in our lab.

In practice, if you take bacterial samples and measure them in different instruments you will IN GENERAL find quite significant disagreement. E.g. even between plate readers of different manufactures, plate reader vs cuvette-based OD meter, or nanodrop type methods, etc. This is particularly true at “High” ODs (i.e. >0.1 or so) which is the multiple-scattering regime where light may bounce of multiple particles (cells) when passing through the sample.

So as a conclusion – IMO I wouldn’t worry about it. There are big differences between machines and measurement modalities even if they “claim” to be doing the same thing. If you are really caring about such things (e.g. the differences between OD = 0.35 and OD = 0.5) because your science requires it then you will need to be very careful and make standard calibration solutions and collect a range of concentrations across all machines and then use this to convert between them. But even then – I think you will be lulling yourself into a false sense of security since there will be still batch- and time-variation in all these measurements.

[harrison]

Hello,
In essence you could make it arbitrarily slow by (for example) turning it on only once every 10 minutes, 1 hour, day, year…. so it is difficult to answer.

In practice you can do somethin glike 0.5ml/minute just using the integrated once-per-minute turning it on for 0.5 seconds every minute.
If you wanted to go slower than this you could make it do one aditionl every 10 minutes, so it would be 0.05ml/minute but delivered at 10 minute intervals.
You can extend this out indefinitely but the caveat is you end up with fairly irregular dosing (i.e. it is NOT continually feeding liquid but at a very slow rate)
hope this makes sense

[Author]

Posts