January 17, 2016 at 01:29AM

Today I Learned: 1) ...one simple technique for detecting manipulation of JPEGs called error level analysis (ELA). ELA is based on the fact that JPEGs (and JPGs) are compressed in a lossy way -- every time you resave a JPEG, it loses some information. To perform ELA, you intentionally resave the JPEG at a relatively low quality level, then compare the result to the original image. You can't usually see the difference by eye, but if you essentially subtract off the lossily compressed image, you can get a sense of how much compression the original image already had gone through -- if the ELA image comes out dark, it means the image was already quite compressed, so resaving it didn't do much to change it; if the ELA comes out bright, then the image changed a lot when it was compressed, so the difference was large. In any case, in a typical image, an ELA is more or less even (though usually salt-and-peppery) across the image (with some exceptions like patches of solid color, which compress very cleanly and so always come out black in the ELA). If part of the image was photoshopped, that part will usually have a different level of compression than the rest of the image, which can be seen in the ELA as a patch of wildly different intesity from the original image. How can you fool an ELA? I imagine you could just resave a JPEG a bunch of times to bring everything down to a low quality level. So now you know to be a bit more suspicious of low-quality JPEGs.... Thanks to Chris Lennox on this one! 2) DNA origami* is one of the coolest techniques on the planet, and one of the things I'm keeping my eye on to potentially radically change medicine, but it has a glaring flaw -- DNA origami structures really don't do well in physiological conditions. There are two main reasons DNA origami breaks down in vivo. The most-cited problem (the one I knew about) is that DNAses** tend to chew them up pretty rapidly; the other main problem, which I learned today, is that living bodies have much lower magnesium concentrations than the folding buffers usually used to make DNA origami. The magnesium ions are important for making the DNA "sticky"; at lower concentrations, the origami gets floppy and falls apart. Today I learned a technique for fixing the second problem and at least somewhat mitigating the first. The key idea was to supply another source of positive charges in the form of oligolysine, which is a bunch of the amino acid lysine strung together (lysine is positively charged and realtively small and therefore unlikely to get too much in the way). If you supply oligolysine at the right concentration, it stabilizes the origami. You still have to *fold* the origami in a magnesium buffer, but then you can add oligolysine, move the origami to living tissue, and it will be fine. Oligolysine also protects modestly against DNAses, probably just by getting in the way. You can further protect origami against DNAses with polyethylene glycol, or PEG, which is usually used in biology as a "crowding agent", or a chemical that takes up a bunch of space to effectivley make other things act like they're at higher concentration than they are. PEG also seems to get in the way of DNAses, and can be incorporated into origami by chemically linking it to oligolysines before adding them to the origami. Adding PEG-K10 (PEG attached to a 10-lysine oligo) extends the half-life of (some particular) DNA origami in physiological conditions (i.e., in cells) from about 3 minutes to about a day. That's a good start. *I've written about DNA origami before, so I won't go into too much detail, but basically it's a technique for folding a very large circle of single-stranded DNA into a stable shape of your choosing. It's ridiculously easy, remarkeably robust, and shockingly close to being cheap (it's about $1,000 worth of DNA to make one). ** DNAse = an enzyme that breaks down rogue DNA. 3) Yeast cells spend part of their life cycle as a haploid* and part as a diploid**. A diploid yeast can undergo one of the several kinds of fission yeast can undergo, resulting in two daughter cells with one set of chromosomes each. Depending on which version of chromosome 3 each daughter gets, it will either be a MAT-A or a MAT-α cell. MAT-A and MAT-α are a bit like male and female, except that one doesn't have a significantly larger reproductive cell than the other. When, however, a MAT-A and a MAT-α yeast come together, however, they fuse, combine chromosomes, and become a single diploid yeast cell. * haploid = has one set of chromosomes, like a sperm or egg cell. ** diploid = has two sets of chromosomes, like almost any cell in an adult human. 4) Bonus: I heard a claim (from someone much smarter and more experienced than I) that the largest known "atomically precise" biological structure is the nuclear pore complex. I'm unable to find a side-by-side to-scale comparison of the nuclear pore complex and, say, the ribosome, but I saw one earlier today and I was very impressed by the nuclear pore complex.