I have tried to use both the Raspberry Pi better than android mini PC. I wanted to use small embedded PCs to run little instruments in the lab. The Raspberry pi is my preferred option.Ii has a good community of users, an easy to install linux OS on the microSD card, and more USB slots for things like the 3D printer. I used it with pronterface to print things and it worked out well.
The android miniPC (mk809IV) turned out to be a bit of a flop. It’s hard to install linux. The native OS it slow. It will install android apps like netflix, which is cool for a home entertainment scenario, but is really too slow even for that.
I was thinking about the value of entertainment. Value is subjective. Someone got paid because Hanson’s 1997 album sold well. Someone got to put money in their retirement fund because that band was successful. There’s nothing wrong with acknowledging the value of pop culture.
Science happened in 1997 as well. The Clustal algorithm (or one iteration of it) was published in 1997. I was using the Clustal algorithm last week to look at my sequencing data. The first humanized monoclonal antibody was FDA approved in 1997. Humanized antibodies have become a pretty big deal.
It’s interesting to compare those two cases: the algorithm is freely available; humanized monoclonal antibodies are rather expensive drugs. One has a clear price tag, but both have clear value. Indeed, the crustal algorithm (or some similar variant algorithm) was almost certainly used to process the data that ultimately led to some of those monoclonal antibody therapeutics.
So here we have three cases: a freely available algorithm, an expensive drug, and a very successful pop culture phenomenon (MMMBOP!) all debuted in 1997. Do they not all have value?
This all leads me to an argument for government-sponsored science. These three phenomena needed different funding mechanisms. Market-based funding worked for Hanson and for humanized monoclonal antibodies; it did not work for the Clustal algorithm. It would be very unfortunate to lose any of this value just because of a commitment to market-based funding.
A new paper in the IEEE MEMS Journal talks about a single chip atomic force microscope. Typical atomic force microscopes run anywhere from $10k to $200k. Atomic force microscopes are very expensive because they require complex alignment of an extremely small pointed object (like a little miniature stylus) to the thing you’re trying to take a picture of. They work by scanning that sharp tip across the object like someone reading braille. The result is that you get a picture of that object that is not limited by the wavelength of light (it’s hard to take pictures of things that are smaller than the light itself).
Anyway, if you can make an atomic force microscope on a single chip, you don’t have to align anything. It just is all built into the same precisely manufactured device. It might be perfect for looking at DNA nanotechnology. I wonder if it would be possible to integrate that with a optical microscope and maybe a laser trap. That would be really good for looking at cells. Or even microparticles like I like to make. It’s probably not going to be good enough to look at single organic molecules (as cool as that is).
Raman spectroscopy has been used to analyze beer before. There’s even a undergraduate exercise that will let a student calculate the percent alcohol in a sample. I want to try it. It’s definitely not just an excuse to try a bunch of different beers.
Raman is like the complement of IR spectroscopy. People are familiar with IR spectroscopy because it is commonly used in organic chemistry classes. Raman is better for aqueous samples than IR (usually) so it makes sense to apply it to beer. This spectrum adapted from Zou et al shows why. The peaks for ethanol are nicely distinguishable above a low background from water.
There are nice hand-held Raman instruments that will give all kind of useful information about a unknown substance. It has been used for things like law enforcement (is that powdered sugar or freebased adrenal glands?) and Department of Defense business. So it’s natural to apply it to beer. It should be possible to amass a big database of beer spectra, annotate it, and mine it for useful information.
But what we need is a cheap instrument that beer enthusiasts can use to take spectra and contribute. A Beer-o-scope or Beercorder. I’m not sure about the name. But as fun as that sounds, I do need to make sure it will work. To the lab!
“The wheels of open-source,” is a plea to not keep reinventing the wheel. It resonated with me because I love reinventing wheels. I’ve been writing code since 1997. My favorite thing about writing code is implementing my own idea and seeing it work. It’s addictive. The fact is that someone else had that idea, implemented it first, then tested it and improved it. Using their code would have huge advantages, but I don’t get the same dopamine rush.
Logic and preference diverge.
I’ve been learning to use BioPython to analyze high throughput sequencing data. I’ve tried k-mer analysis based on an article from 2011, and that sort-of worked. I got an aptamer… but it didn’t bind native protein, only recombinant protein. And it was the most successful candidate that came out of the k-mer analysis.
So: back to the bench, or back to bioinformatics?
Might as well try to get more from the data we have, I thought. The trouble was that… bioinformatics is hard. I could figure out how to do k-mer analysis by trial and error. How about phylogenetic trees? How about a scoring system for sequence similarity? These things have been done carefully and checked for statistical correctness and computational efficiency.
So… it’s time to not reinvent the wheel. It’s time to follow logic.