Pioreactor development log #5

Pioreactor development log #5
One of our goals with the Pioreactor is design it such that you don't need to be a biologist,  or an electrical engineer, or relevant for this article: a statistician. This article describes our internal algorithm that computes the culture's growth rate, but importantly: you don't need to know this algorithm to use the Pioreactor! We've designed the internal statistical algorithm to be robust enough that you can sit back and watch. This article is for the users who really want to dig deep into how we compute growth rates and the statistics behind it. 

Replication, variance, (and growth rates) in Pioreactors

Replication, variance, (and growth rates) in Pioreactors

One important property we wanted Pioreactors to have was low variance between units. That is, one Pioreactor should be interchangeable with another. This property means that differences in experiment results are solely the result of biological culture conditions, and not on the Pioreactor construction or materials. In fact, we take inspiration from cloud computing providers: the customer shouldn't have to worry about which hardware they are running on, and the cloud provider can swap out hardware without the customer noticing. 

Low variance between Pioreactor is accomplished through hardware and software design. The hardware is designed such that outputs, like LEDs, try to be as consistent as possible. However, there are things that will always be out of our control: imperfections in the glass vial, variation in commodity hardware components, slight misalignment of LEDs, etc. To help fix these problems, our software performs normalization tasks that try to control for variations. After all this, we are really happy with how consistent the Pioreactors are! 

As an at-home test, we prepared a very simple YPD broth, let cool, added a small amount of Baker's yeast and gently stirred. We added about 12ml of the broth to three Pioreactors and let it grow for ~18h. We expected to see very similar growth behaviour between all three Pioreactors since the cultures were identical. Below are the results of the optical density (our measure of culture size):

Optical density of growing yeast, over time. Experiment is a triplicate with two sensors in each Pioreactor.

Each Pioreactor had two sensors, hence the six lines. We can see a high alignment between Pioreactors - each growing at similar rates (we'll see this is more detail later), and each hitting the growth plateau at a remarkably close time (this is when the glucose was exhausted). 

The above graph is tracking optical density, but Pioreactors go one step further and infer a real-time growth rate as well. This is very useful, as growth rate is not constant over time, and varies quite a bit. And knowing the growth rate is key. In fact, understanding and controlling growth are fundamental to understanding and controlling microbes:

The study of the growth of bacterial cultures does not constitute a specialized subject or branch of research: it is the basic method of Microbiology.

- Jacques Monod, 1949, biochemist and later Nobel Prize winner 

Aside: a goal of Pioreactor is to make growth rates as accessible as possible. You don't need a degree in biochemistry or statistics - we make it a basic method for you. 

Below is the growth rate for our three cultures: 

The implied growth rates of the yeast cultures.

 

First thing to notice: we again see excellent alignment of growth rates. There is little variance in the three cultures. Next thing to notice are the subtleties in growth rate!  The growth rates increase and plateau, increase again, fall, etc. These changes are not at all obvious when looking at the optical density alone. I hope Monod would be happy to see such highly detailed, and consistent, growth curves. Only 70 years later! 

Pioreactor will only get better from here, and we expect more and more variation to be controlled, either in hardware or software. You, the curious biologist, get the benefits, and a new tool in your toolbelt. 

Don't be fooled: Yeast have two growth spurts!

Don't be fooled: Yeast have two growth spurts!

Did you know that the traditional S-shaped growth curve in microbiology is terribly inaccurate? It's inaccurate for many reasons, but the reason I want to discuss today is that some microbes have multiple growth spurts - like two "S"s ontop of each other. For example, when there are two sources of sugar in the media, the microbes will first consume the more accessible sugar, and then after a pause to reconfigure their metabolism, and then start to consume the second source of sugar. This leads to two growth spurts, and hence two "S"s. 

Interestingly, yeast have another trick up their sleeves (cell membranes?). The consumption of sugar, like glucose, gives yeast a large initial spurt of growth, as you can see in the below figure (the output of our Pioreactor). As a byproduct, the yeast expel alcohol, specifically ethanol. 

After this consumption of sugar, the yeast are not done. After all the sugars are consumed, they take some time, reconfigure their internal metabolism, and start to consume the ethanol they just expelled! 
Let's zoom out further:
In the above, we can see this new "double-S" growth of the yeast culture. But this new growth is a lot less pronounced in the implied growth rate chart. Why is this? When the yeast consume the available sugar, the culture 5x its initial size (starts at 1, and goes to ~5 after 3 days). That implies a very high growth rate. However, when the yeast consume the ethanol, which has less available energy in it, the yeast are only able to 2x their population size (from ~5 to ~10). Hence the implied growth rate is much smaller and flatter. 

This phenomenon is called the diauxic shift, and only some yeast and bacteria can achieve it. The ethanol is oxidized, which means that oxygen is needed for this second-stage of growth to occur. 

The above experiment was performed using our Pioreactor. The Pioreactor can help you study the impact of media, temperature, and other variables on the diauxic shift. Have fun!