This looks along the same lines as CyberEther, though CyberEther does have alot more features to enable massive compute of raw data and arguably nicer gui.(https://github.com/luigifcruz/CyberEther)
The best articles for this are [1] (google translate it) and [2]. In [1], the author uses .asc files while in [2], .img files are used. You can treat these files as basically the same as .hgt but remember to reproject the data as different DEM sources uses slightly different projections.
Not to be confused with citizenlab.ca, a research lab of University of Toronto that deals with high profile civil right violations using OSINT and tech analysis.
They are focused on completely different things though, this seems to be neighbourhood projects in Belgium and the Netherlands. Not focused on privacy and citizen data at all. But yeah they should have picked a different name.
This is not correct. The security page only shows that they use Google for Speech to Text.
Instead, Text to Speech is done using technology developed by Lyrebird.ai, which Descript has bought over. Descript rebranded it as Overdub. Note that style transfer learning of voices is a hard problem and Overdub seems to have nailed it perfectly. I speculate that the underlying technology of Overdub is based on sv2tts (https://arxiv.org/abs/1806.04558).
Kudos for the fascinating article. Since you mentioned security, and the well known trick of sending an ambiguous message that is interpreted differently by different interpreters, I wonder what you might think of enlisting the hackers of the world to help in viral research. Seems as though adversarial thinking skills could come in handy?
I definitely think the computing / information security mindset is relevant and applicable to computational biology. There are sadly structural reasons why the crossover is very hard. I spent 18 months at a university doing this kind of research but it is not a natural fit. Also, I do have to tell you, I spent 10 years studying this stuff before I became useful to the field. So you don't just jump in there :-)
Thank you for the thoughtful reply. You read my mind.
I would be happy to devote 10 years (or a lifetime for that matter) to bring the fields to bear on one another, sounds super interesting.
As others have already noted, most work at the "intersection" of the domains is currently superficial, say designing support software as opposed to actually hacking on the genetic code.
Could you go into more detail about the structural reasons that prevent a clean mapping? Can you think of ways to lower the barrier so as to engage more people? Do you see problems that are relatively easy to export?
An example of exporting problems, in this case from digital pathology, is the Camelyon challenge, which brought machine learning to bear on cancer detection. Hype aside, the important aspect is that researchers working to segment images and train neural networks did not have to have a biochemistry background, nor understand cellular functions and provenance, nor understand staining protocols. A clean export.
I am basically looking for a "biocrackme" that is relatively self contained.
yeah, I'd love to know as well. Both Moderna and BioNTech licensed technology from University of Pennsylvania, but I don't know if that is only the 1-methyl-3’-pseudouridylyl bit. If I find out I'll update the post.
