iskander

Illumina error rates are <<1% (~0.1%), whereas Nanopore with newer basecalling software is 5-10%. With UMIs you can get a consensus error that's also <<1%. The error profiles are indeed different: Illumina generally creates substitution errors, whereas Nanopore has trouble with "homopolymers" -- counting how many of the same letter occur in a row.

No true Redstate?

Setting aside blue states with moderate Republican governors, I see: 1/10 (Duke) 5/20 (Duke, Notre Dame, Vanderbilt, Rice, Wash U) 15/45 (Duke, Notre Dame, Vanderbilt, Rice, Wash U, Emory, Ann Arbor, Wake Forest, UVa, Georgia Tech, UNC-CH, UF, Case Western, College of William and Mary, Tulane)

I doubt the split would be at the top 10 because they're simply too small to absorb all the domestic demand. I'm not sure how far down the list, however, would be the switch point between benefit and harm (or if it would be determined purely by rankings vs. some other consideration).

That's not how Shabbat works...

(it's very mysterious that someone would go to all this trouble with an idiosyncratic outsider interpretation of Judaism)

Peptide vaccine for SARSCoV2

I just started working on the complementary side of the spike RBD-ACE2 interaction, making an aerosol of RBD peptide fragments to outcompete the virus for entry.

This work is based on B-cell epitope mapping of SARS patients and subsequent identification of ACE2 binding peptides like: https://www.sciencedirect.com/science/article/pii/S016635421...

I'm the "dry lab" side of this project -- looking for peptide candidates. My other collaborators at UNC are going to do formulate the aerosol and test for safety in mice.

Normally this would be a very early stage in drug development, but it seems like we have to move quickly.

So similar shout out to the Gates Foundation folks -- if you're interested in this approach email alex.rubinsteyn@unc.edu

Detailed rebuttal: https://docs.google.com/document/d/1niBInt7tgQ57iihgenPFMyn7...

tl;dr this paper should be retracted

I'm going to sidestep arguments about its terseness since some people are into that and it's really an aesthetic choice. But for machine learning or numerical algorithms, I found it quite hard to make k nearly as fast as vanilla NumPy code. Overall there's really no good reason for it to be faster. It didn't have (as of ~2011) a JIT or any mechanism for operator fusion, so you end up allocating many intermediates, sometimes using quadratic space which gets immediately reduced along some axis.

Yeah, I know "Arthur Whitney is really smart"...but that's not actually a technical reason why k would be faster. I don't remember the specifics but I think that many of the array operators weren't even multi-threaded and I remember only a few of the array operators getting FLOPs that would correspond to SIMD acceleration. So, for the most part, it seemed like a lot of vanilla single threaded C code behind the operators.

It's possible that the implementation got smarter over the past decade, but when I was working in the space it seemed like the hagiography and actual performance were worlds apart.

>There are hundreds of benchmarks for kdb+, and the speed of k is unrivaled.

Eh, I worked on array languages in grad school and the speed of k was significantly over-hyped. It's trivial to beet with Python and no contest if you want to work in a compiled language. Wall Street develops weird religions around tech.

I grew up on the North side of Chicago and also didn't feel like UChicago had a very large presence. Now that I work in the research world (both in CS and biomedicine), it's very obviously the dominant institution in Chicago. Northwestern comes in at a distant second and UIC isn't really in the race. I'm not sure about their undergrad education but if a student wants to have access to teachers who are also leading voices in their fields, UChicago would be a better bet than other local options. Nationally you would have to look to places like Harvard or Stanford to do better.

I transitioned into cancer immunotherapy research after Computer Science grad school. Biology is a mess and it will take several years to get up to speed on any one research area. The best options are either grad school or finding a job which doesn't initially require a lot of bio background but provides the opportunity to learn. It's also worth checking out local biohacker spaces, they can teach some basic techniques but you'll hit a ceiling quickly. To really "start a biotech company" you'll really need to spend a long time outside your comfort zone.

It seems like it's a commonly used undergrad textbook?

The average lifetime risk of getting cancer in the US has actually fallen substantially due to a decrease in smoking.

Diagnosis has simultaneously increased, especially in cancers where the rationale for treatment is low (e.g. prostate cancer in most men).

(work in cancer research)

I like https://www.mixily.com: it's not a direct Meetup competitor but it is nice for scheduling individual events.

I've been using Mixily (http://www.mixily.com) for events with friends and would happily use it to schedule individual meetup events. It's missing a main community landing page though.

:-)

I have a dog named Laika.

I wouldn't be so sure, a cancer that doesn't express classical MHC but still manages to suppress NK cells has a chance. It would probably have to master a few other immunosuppressive pathways but it's not beyond imagination.

Transplanted HeLa cells were shown to engraft in unrelated human "subjects" (in some truly evil experiments by https://en.wikipedia.org/wiki/Chester_M._Southam), other cell lines probably have the same capacity.

Coolest finding in the paper:

"Our analyses reveal a mutational signature, signature A, that occurred in the past but ceased to be active from about 1000 years ago. A recent study (37) detected evidence for an excess of C>T mutations at TCC contexts, the mutation type most prevalent in signature A, accumulating in the human germ line between 15,000 and 2000 years ago. If this human mutation pulse is due to signature A, it could indicate a shared environmental exposure that was once widespread but has now disappeared."

There will now be a small academic race to identify this mysterious carcinogen which both humans and this dog cancer were previously exposed to.

I don't know about generalizations across all "modern historians" but Biblical Minimalism (https://en.wikipedia.org/wiki/Biblical_minimalism) was a big trend in near Eastern archaeology. They had/have legit criticism of how modern definitions/attitudes taint our interpretation of the archaeological record.

But...ultimately, that view ignores that many of the places (and even some of the people) of the Biblical narrative from the 9th century BCE seem to roughly match the archeological record.

And even before that the record is simply incomplete or mixed, rather than directly contradicting the Biblical era of prophets and early kings.

For example, at the Elah Fortress site (10th century BC), there were no pig bones (unlike Philistine sites) and a fragmentary inscription that at least resembles the Israelite culture depicted by the Bible:

1 you shall not do [it], but worship (the god) [El] 2 Judge the sla[ve] and the wid[ow] / Judge the orph[an] 3 [and] the stranger. [Pl]ead for the infant / plead for the po[or and] 4 the widow. Rehabilitate [the poor] at the hands of the king 5 Protect the po[or and] the slave / [supp]ort the stranger.

This isn't evidence for the Bible (which makes lots of other unprovable magical claims) but rather evidence that parts of the Bible do accurately convey information about an ancient society.

The most common belief is that RNA came first -- it has both the information-carrying capacity and enzymatic activity.

As for why are there 20 encoded amino acids (+ selenocysteine and pyrrolysine)?

I've read that codon similarities hint at an original more compact alphabet of two base codons (allowing for up to 15 distinct amino acids and a stop).

I've also read that the "more recent" amino acids, like cysteine, are more readily oxidized and thus provided some advantage as atmospheric oxygen levels increased.

But all of these are guesses about the world before LUCA, which is really lost to us.

These evolutionary backsplanations feel like they're saying more about our current society, values, and biases than any group of past humans.

Maybe helpful for the HN crowd: the adaptive immune response typically prioritizes a low false positive rate. A developing cancer sneaks by as a false negative. Checkpoint blockade moves the immune system to a regime of higher sensitivity, at the expense of potentially lower specificity.

The immune system is thought to already clear mutated and defective cells as part of their everyday activity.

Cancers which grow to a detectable/dangerous population have to find some way to evade this immune surveillance and hitting off switches like "immune checkpoints" is one way to accomplish this. So, blocking the off switches helps the immune system resume its everyday task of killing weird cells.

Surreal to be at the CRI / cancer immunotherapy conference while this got announced.

I currently live split between NY and Miami but am starting to look at jobs outside the US.

Meh. I've spent enough time in the Netherlands and Israel to know that the quality of life in both countries is, in many ways, way better than here. Socialized medicine and cheap education go a really long way toward making people happy (even if they have less disposable income).

It's worth reading the article, it's about an interesting study in lab mice.

No opinions on how anyone should update firmware but I just wanted to chime in that I bought a fairly expensive System76 laptop in 2011 and was very disappointed with how rough & clunky it was.

I deleted Uber during that campaign and so did at least a dozen people that I know. I don't know the exact numbers, but I think Lyft's market share roughly doubled within several months.

How did "the market" not notice?

>And I'm pretty confident the majority of experiments are using sample sizes larger than 12.

It definitely varies by field, testing 3 doses and a negative control with 3x replication would be a pretty good experiment for testing a drug in mice. It's plenty of data for identifying a drug that actually works, as long as your statistical tools aren't designed with expectation of thousands of samples.