Week 1 HW: Final Project Ideas

Question 1

I would love to biomanufacture a living diagnostic system inspired by Baymax from Big Hero 6. My ideal microorganism would be capable of sensing physiological signals associated with disease and converting them into detectable output, for example, into a low-background VOC that can be sampled non-invasively (via breath or urine)

My motivation for the idea comes from the assumption that biological systems can process information about the body in ways that technology cannot necessarily replicate. As an example, maternal milk dynamically changes its composition as a response to the needs of her infant. My belief in this idea’s importance comes from the countless medical conditions that go undiagnosed or dismissed due to lack of distinct and straightforward tests. My long-term goal is to explore whether engineered cells could become biological Baymaxes that can just as in the cartoon detect diseases.

Question 2

One method that I thought would be fun to explore is to have the engineered cell record the vitals and/or internal environment of a healthy body, and when assigned to a sick person (with an assumed undiagnosed condition), compare the two environments. That way, we could study what metrics differ in the sick person and find the cause of sickness. For this method, we might need a very sensitive chassis with high precision.

I wonder if it is possible to code my engineered cell to synthesize a molecule as it reads through the healthy body, in a way encoding the inner environment. When put into a sick body, have the synthesized molecule unravel if all the metrics match and stop the travel through the body at a mismatch. Assuming that the cell will take the same pathway when travelling, the stop point is the one that has an issue. The most suitable chassis so far would be E. coli, with a modified protein synthesis apparatus.

Question 3

I would like to prototype a cell that will follow a foreign line and remember it as its baseline, then follow a line that starts off the same way and deviates at one point, and check whether the cell would stop. I think high-throughput automation could be used to find the most perfect chassis (maybe better than E. coli) or a genetic circuit, by trying running numerous tests with different molecules.

Question 4

a) If this engineered cell would act like a one-time testing kit, some solid medical data can be collected from a sample of 1000 tests. A question would be how many engineered cells one test will require. Because the number of cells required for a reliable diagnostic measurement is one of the major unknowns in this idea, I will treat it as an engineering parameter to be experimentally determined. For the initial manufacturing estimate, I assume 10^{7 cells per test and would later optimize this dose by measuring the relationship between cell number and probability of obtaining a complete molecular record. For the batch of 1000 tests, then, we would need 10 billion engineered cells. With the help of AI, I discovered that these cells come in a culture volume of approximately 10}9 cells/mL. That would mean that 10 mL of sufficiently dense culture could contain enough cells for 1000 tests. However, to account for losses during harvesting, quality control, and failed batches, we can put 10x manufacturing buffer on it, and end up with 100 mL to start.

b) The development part for selecting the final E. coli strain might require a lot of experiments, while during manufacturing, 1 ready strain will take 34 cycles to multiply into 10 billion cells, which is around 11 hours

Question 5

I need to determine whether the engineered cell can distinguish disease-related physiological differences from normal variation between individuals. For example, differences in genetics, metabolism, or other individual characteristics should not be mistakenly interpreted as signs of disease.