Craig Venter was famously involved in the Human Genome Project. He announced the first draft of the human genome alongside President Clinton and Francis Collins.
"Involved" in the sense that he took the public data, added in a small amount of his own privately generated data and was trying to get the first assembly. The scientists in the Human Genome Project thought he was going to try to patent the whole thing so others would have to pay him. Back then, it was not clear what was and was not patentable.
So the involvement was in spurring the Human Genome Project to race to an assembly, a massive computational problem that hadn't been fully planned for by the public effort:
It was essentially a jigsaw puzzle, and Venters insight was that computational power was just as important to the project as biology. The Human Genome Project was essentially trying to sequence the human genome by finding large chunks of DNA and fitting them together like a jigsaw, finding which bits unambiguously matched up.
Venters idea was that you could do the same with small chunks of DNA, if you approached it as a computational problem and used computers to try/evaluate/reject the millions of ways the pieces could be fit together. So he recruited mathematicians, computer scientists etc and got them to work on the problem. He speeded the project up massively by making the biology bits simpler (smaller pieces of DNA) and shifting the effort to the computational problem.
So he made a big difference. And his insight that it was a computational problem is kindof obvious now but it wasn't obvious 25 years ago.
It was very obvious that it was a computational problem, all DNA analysis was highly computational then, as it is now. His guess was that ~500bp fragments would be enough to get a usable assembly.
But the Human Genome Project's approach of reconstructing larger chunks first was also feasible, and produced an assembly too, with a heroic four weeks effort of a former game programmer who even built cluster software at the same time.
He wasn't the only one who saw the problem computationally. Famously, the mathematician Michael Waterman sat on the other-side of the race for the human genome.
Involved in the sense that his method worked and the one the Human Genome Project insisted on didn't. In the end, they had to use his method to catch up enough that everybody could pretend they did it together and collaboratively -- even though Venter clearly got there first. Venter deserved a Nobel Prize for that and, quite frankly, the Human Genome Project guys deserved a firing.
Craig Venter had his genome sequenced in 2007. It was the first individual human genome that was sequenced and released publicly.
The human reference genome is ~70% from a man with African and European ancestry who lived somewhere around Buffalo, NY. Most of the rest is from ~20 other individuals in the same area. They were supposed to sequence the samples more evenly, but apparently there were some technical reasons that made them prioritize a single sample.
I worked on this back in the 90s and there multiple data sets being used. We had one that was Mennonite family with like 5 living generations and 100ish individuals.
It gets a little fuzzy when talking about Celera and the human genome project. The two efforts were very much competitors, but there was a lot of crossover (mainly from Celera pulling in the public data).
But, Venter claimed that he was the a good chunk of the genome that Celera sequenced, so I think it's fair to say he was one of the people included in the draft human genome (at least the Celera version of it).
> After leaving Celera in 2002, Venter announced that much of the genome that had been sequenced there was his own. [1]
I am not sure what is "the draft human genome" you are talking about. Two separate human genomes were published in 2001: the HGP genome and the celera genome. The HGP genome then didn't use Venter DNA. It evolved into the current human reference genome. The celera genome contained Venter DNA but it has been completely forgotten nowadays.
> That said, using a vintage technology has some downsides. It was never updated to support WebSockets
With widespread browser support for WHATWG streams, it's pretty easy to implement your own WebSockets over long-lived HTTP requests. Basically you just send a byte stream and prepend each message with a header, which can just be a size in many cases.
Advantages over WebSockets:
* No special path in your server layer like you need for WebSocket.
* Backpressure
* You get to take advantage of HTTP/2/3 improvements for free
* Lower framing overhead
Unfortunately AFAIK it's still not supported to still be streaming your request body while receiving the response, so you need a pair of requests for full bidirectional streaming.
Not the same. SSE doesn't support binary, and has higher framing overhead than you can achieve with raw HTTP. If you don't need binary it's a solid option in the browser.
It's useful as a rough heuristic, but tends to overestimate utilization. We've also noticed that power-derived metrics have a lag time behind true utilization, the controller that regulates it has a delayed response time. This especially becomes important for spiky workloads like real-time inference.
Any tool (like nvtop) that only queries NVIDIA's NVML library does not have access to the detailed metrics that we draw upon, and therefore has to use proxies for efficiency.
Crazy seeing this posted today. I learned about flipdiscs from ChatGPT just a few hours ago because I'm trying to find a text display type you could build or at least repair yourself. Don't need much just ASCII with enough lines for running commands and output.
I think flipdiscs would be too expensive. Honestly a teletypewriter might be the most realistic, but wastes a lot of paper.
reply