Machine learning meets microgravity genomics
Source: NASA Apollo Programme Photo Archive
Whenever I start writing a research paper, the famous aphorism by Steve Jobs comes to my mind.
‘Again, you can’t connect the dots looking forward; you can only connect them looking backwards.‘— Steve Jobs
As much as I dislike embedding aphorisms in my research work, that simple quote pretty much summarizes for me how machine learning will tackle the most of the existing problems in protein biotechnology.
I don’t bet my money on bizarrely complex deep learning. I put it completely on the experimental data, along with a deep and profound understanding of their type and nature. Why? Because numerical machines built to predict and explain natural phenomena are as good as your input data are.
You may wonder now, how is my little machine learning digression related to space poop in the first place?
NASA have organized their bizarrely awesome Space Poop Challenge. Beyond the obvious arguments, space poop-related research can have an invaluable impact on the modern biotechnology via machine learning. Allow me to explain why.
But first. Why pooping in outer space is a bad idea?
1. It’s cold outside.
The current black body temperature of the background radiation is about:
3K = −270°C = −454 °F
Bad for your bottom. On the other hand, awesome for quantum effects enthusiasts, who would love to experiment with superconducting and superfluidity.
2. It’s dark.
With an average distance between starts of 4.2 light years, one can safely conclude outer space is quite dark. Moreover, if our expansion models are correct, it is about to get even darker and colder.
3.It smells funny.
As Don Pettit, interestingly explains, space does smell funny.
It is hard to describe this smell; it is definitely not the olfactory equivalent to describing the palette sensations of some new food as “tastes like chicken.” The best description I can come up with is metallic; a rather pleasant sweet metallic sensation. It reminded me of my college summers where I labored for many hours with an arc welding torch repairing heavy equipment for a small logging outfit. It reminded me of pleasant sweet smelling welding fumes. That is the smell of space.
4. You may get hit by something nasty.
Don’t take my word for it. Simply check NASA Frontier Development Lab. Our dearly beloved Earth is exposed not only to abundant space debris. There are all sorts of celestial objects, which require constant monitoring and completely unorthodox detection methods.
NASA Frontier Development Lab_Edit description_www.frontierdevelopmentlab.org
5. Exo- and astrobiologists will hate you forever.
You may want to note this one down. A gut of an healthy individual can host on average 390000000000000 bacterial cells, which means you carry with you almost an equal amount of bacteria as the number of the cells that build your body! A human stool contains on average from 540000000000 to 720000000000 bacterial cells per a gram of dry weight!
Interestingly enough, common bacteria are known to display bewildering resistance to cosmic radiation, oxygen shortage and even stringent sterilization. Let me quote an abstract of one of the seminal studies of bacterial survival rate in space.
On board of the NASA Long Duration Exposure Facility (LDEF), spores of Bacillus subtilis in monolayers (10⁶/sample) or multilayers (10⁸/sample) were exposed to the space environment for nearly six years and their survival was analyzed after retrieval. The response to space parameters, such as vacuum (10^(-6) Pa), solar electromagnetic radiation up to the highly energetic vacuum-ultraviolet range (10⁹ J/m2) and/or cosmic radiation (4.8 Gy), was studied and compared to the results of a simultaneously running ground control experiment. If shielded against solar ultraviolet (UV)-radiation, up to 80 % of spores in multilayers survive in space. Solar UV-radiation, being the most deleterious parameter of space, reduces survival by 4 orders of magnitude or more. However, up to 10⁴ viable spores were still recovered, even in completely unprotected samples. Substances, such as glucose or buffer salts serve as chemical protectants. With this 6 year study in space, experimental data are provided to the discussion on the likelihood of “Panspermia”.
Dumping human feces into the lower orbit of earth could create a microbiologically contaminated zone with decremental and unpredictable consequences for all sorts of extracelestial life search missions. Imagine the headlines of science and news outlets: ‘Extracelestial life found!’, ‘We are not alone’, and finally the rectifications — ‘We are alone! We have found space poop’. There is nothing worse in science than false positive results!
6. WYPInWYG — What You Poop Is not What You Get!
This is my final point. It makes me truly giggle with, both excitement, and angst. Common bacteria mutate in microgravity and outer space environment at least order of magnitude faster than on earth.
Why should anyone care about this?
First of all, it appears microgravity, cosmic radiation, and a whole range of unreproducible (on earth) conditions specific to outer space simply fast-forward evolutionary-like mechanisms on the molecular level.
Having an information about what a certain pathogen could become in the near future, is of central interest in industrial and medical biotechnology. We are given a unique chance to prepare an adequate response, should the mutagenesis and epigenetics scenarios observed in microgravity take place on earth.
The experimental data from the ‘fast-forward’ molecular evolution and selection experiments not only enrich our current knowledge about systems microbiology, but can be of fundamental role in the understanding how life evolves on a molecular level under conditions very different from the habitat we have on earth. Thus, we should be able to refine our methods aimed at finding extracelestial life.
Finally, at Peptone — The Protein Intelligence Company, we are actively looking for a synthesis of data from systems microbiology, high throughput genomics and proteomics analyses, to aid our academic collaborators and industrial clients at designing better proteins. I will refer you here to an excellent article written by Matthew Heberling, who in very intelligible way enlists and explains the nature of the biggest practical issues which hamper current biotechnology research.
Fame and Fundamentals Complicate Protein Biotech_Protein Science Deserves More from Big Data_blog.peptone.io
As I have postulated in my previous article, machine learning alone will absolutely NOT solve all the problems we have in protein biotechnology. The key is the data, and the way we ingest them in the context of systems biology.
To sum up: pooping in space may have absolutely unforeseen consequences, which cannot be modeled nor predicted by GANs, Deep Nets, Convnets, Inception nets and any other **/.*net[s]/g****(**generate your own bizarre and catchy name using this regex).
Finally. If you decide to poop in space, please package it and send it back to earth for further analysis!
Useful post? If so, make my day by clicking the heart below!
About the Author
Kamil Tamiola is the founder and the architect of Peptone — The Protein Intelligence Company. You can connect with him on AngelList, LinkedIn, Twitter, and Researchgate.