Lior Pachter

Worth a read, more in depth analysis of some of these studies: https://liorpachter.wordpress.com/2020/11/17/mathematical-analysis-of-…

Hello! I fear you submitted the wrong link. When I follow it, I get to a sign in page for Wordpress rather than the article you intended, which was presumably this one: https://liorpachter.wordpress.com/2020/11/22/williams-math-professor-i… Maybe resubmit?

A classic from Lior Pachter (not mentioned in the article) from 2014: https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a…

Completely agree with your comment, and here you have some more examples in the Spanish university to support your thesis: - Rafael Luque was suspended of employment for 13 years not because of blatant dishonesty but because he was signing with the King Abdulaziz University as first affiliation and that bothered his university [1]. The KAU is known for buying citations this way [2]. - Another example is Francisco Herrera Triguero [3]. He's been signing with KAU as second affiliation for years and nothing happened to him even when the story about the other guy broke out. So yeah, maybe we don't need more algorithms but more public oversight over this stuff. [1] (Spanish): https://elpais.com/ciencia/2023-03-31/suspendido-de-empleo-y-sueldo-po… [2] https://liorpachter.wordpress.com/2014/10/31/to-some-a-citation-is-wor… [3] (Spanish): https://www.elsaltodiario.com/universidad/papers-y-mas-papers-las-somb…

That 40% number is accurate but misleading. Here's an old prediction of pretty much that result that explains the issue: https://liorpachter.wordpress.com/2013/11/30/23andme-genotypes-are-all… Basically, almost all (~99.9%) of the results of tests like 23andme (SNP arrays) are correct, but because an interesting/scary variant/mutation is about as rare in the population as the test result being wrong for a given site, somewhere around half of those scary results are false positive. If 1000 people are tested for a mutation with this method, 10 will be false positive. The real frequency of mutations vary by site - some will be close to 50%, others at 10e-9, depending on how strongly that variant is selected against, so the most harmful ones are very rare. If you get a result from 23andme that your doctor thinks is worth retesting, it's probably because it's a harmful mutation, which is rare, and ergo there's a fair chance the 23andme result is wrong. So 23andme is almost always right, except when the result might affect your health in a very serious way. I should point out that for late-onset diseases like cancer, heart disease, alzheimer's, etc., there's little natural selection, so pathogenic variants are more common, making it more likely that the 23andme result is correct. It's a simple bayes rule result. It would not be hard for 23andme to give you an estimate of how likely the result is to be a false positive, but maybe there's some regulatory barrier in the way?

>Failure to distribute the data and code not only greatly devalues the contributions of many scientists, it makes replication far more difficult, and opens the door for outright fabrication. Reasons why labs don't produce code: (1) Fear of being scooped; suppose you put in tons of man-hours into a custom population genetics association study for malaria in Africa, you don't want your competitor lab to sequence a bunch samples in SE Asia and run your code as-is and publish a paper when you have the sequencers to do the same. (2) Fear of competitors not being able to replicate stochastic ML results; some machine-learning and artificial neural network papers applied to Biology are stochastic. Past authors have been accused of "cherry-picking" the most "optimistic" runs that show positive results, e.g., (https://liorpachter.wordpress.com/2014/02/11/the-network-nonsense-of-m… (3) Focus on science, not tool production; most labs' focus is to produce publications, not open-source software. User base for scientific software is very niche, unlike Web MVC frameworks; making the payoff calculus of packaging and supporting external users not so great. Furthermore, most labs have custom databases making integrating external software, externalizing internal software difficult.

More recent commentary by a math & bio prof: https://liorpachter.wordpress.com/2014/12/30/the-two-cultures-of-mathe… HN discussion (69 comments) https://news.ycombinator.com/item?id=8819811

There is criticism around ALB. See [1] and specific notes from other network scientists such as Bernardo Huberman (Andrew Huberman [2] father). [1] https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a… [2] https://en.wikipedia.org/wiki/Andrew_Huberman

Related: “The perfect human is Puerto Rican” – https://liorpachter.wordpress.com/2014/12/02/the-perfect-human-is-puer…

Context: https://terrytao.wordpress.com/2018/09/11/on-the-recently-removed-pape… https://liorpachter.wordpress.com/2018/09/17/mathematics-matters/ https://quillette.com/2018/09/07/academic-activists-send-a-published-p… (by one of the paper's authors) http://www.personal.psu.edu/cpc16/politico-email-29nov2018.pdf (mail criticizing Quillette article)

A starting point: https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a…

There is a lot of pushback on Levitt's predictions that keep moving the goalposts. Let's just say his Nobel Prize wasn't in curve fitting. https://liorpachter.wordpress.com/2020/09/21/the-lethal-nonsense-of-mi…

James Watson - who was as close to royalty as you can get in science - worked very long and hard to get himself fired. It wasn’t just ”an interview”, but a lifetime of racism and sexism. Lior Pachter has a handy collection of quotes with links to sources: https://liorpachter.wordpress.com/2018/05/18/james-watson-in-his-own-w…

The question is who is most likely to be right, a nobel prize winning Ph.D. specialized in molecular biology, genetics and zoology, who worked 20 years at Harvard etc, etc....or you. Lior Pachter (who that link was from) is a Professor of Computational Biology at CalTech. His other post about Watson is good too: https://liorpachter.wordpress.com/2014/12/02/the-perfect-human-is-puer… But you don't need to steal a Nobel prize to recognise racism. And yes, "scientific" racism is wrong as in incorrect, as any honest search will show.

You could start with Pachter's sendup (https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a… and the Wikipedia article on Barabasi has a short section on criticisms. I read his work, figured out it was nonsense, and put him on the list of authors whose papers I automatically threw in the garbage. Kim Lewis is another one of those.

The methods weren't "just" morally wrong but a bunch of hokum rationalized as science. Just like yours. Heritability does not mean heredity. Even taste is heritable. Results of the GWAS crowd are so bad that Plomin recently truned to praising his levels of signal for being not worse than social sciences. This is astounding from someone previously selling behavioral genetics as a rigorous science up there with actual genetics. Don't even get me started on IQ https://liorpachter.wordpress.com/2018/06/04/low-iq-scores-predict-exc…

This is not new, at least for biological networks. For example, see this paper by Lima-Mendez and van Helden: The powerful law of the power law and other myths in network biology: http://pubs.rsc.org/en/Content/ArticleLanding/2009/MB/b908681a#!divAbs… and several blog posts by Lior Pachter: https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a…

No I mean this blog post: https://liorpachter.wordpress.com/2017/08/02/how-not-to-perform-a-diff…

'I was wrong' part 1 was a really a discussion about how wrong / fraudulent some authors were who published in Nature in 2004. It looks like it turned into a bit of flame-fest[1], with the original authors chipping in. [1] https://liorpachter.wordpress.com/2015/05/26/pachters-p-value-prize/#c…

I learned about Apostol's death from a post by a very astute mathematical biologist, Lior Pachter, who was once a student of Apostol's. The post is "The Ice Cream Cone Proof,"[1] which nicely illustrates Apostol's brilliance as a teacher. [1] https://liorpachter.wordpress.com/2016/05/09/the-ice-cream-cone-proof/

Be careful about Barabasi: https://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-a… (FWIW, I had previously read some Barabasi papers and had come away seriously unimpressed, see also https://news.ycombinator.com/item?id=9555547)

Lior Pachter on read alignment in general is a good start[1]. This package seems to modify the human reference genome based on the known mutations a person has, making it easier to map an exon back to a genomic position. [1] https://liorpachter.wordpress.com/2015/11/01/what-is-a-read-mapping/

From a slightly different angle, Lior Pachter focuses on how science is often a pyramid scheme of sorts: https://liorpachter.wordpress.com/2016/01/18/the-heroes-of-crispr/ >There is a more cynical view of the PI model, namely that by running large labs PIs are able to benefit from a scientific roulette of their own making. PIs can claim credit for successes while blaming underlings for failures. Even one success can fund numerous failures, and so the wheel spins faster and faster…the PI always wins. >Lander concludes his perspective by writing that “the [CRISPR] narrative…is a wonderful lesson for a young person contemplating a life in science”. That may be true of the CRISPR narrative, but the lesson of his narrative is that young persons should not count on leaders in their field to recognize their work.

I think Bioinformatics is a challenging field because, as the residence alludes to, it is aiming to marry different fields that don't necessarily talk to each other. Lior Pachter has blogged about this [1]. Pachter's book, Algebraic Statistics For Computational Biology is fairly accessible. [1] https://liorpachter.wordpress.com/2014/12/30/the-two-cultures-of-mathe…

Lior Pachter had a blog post about this incident (The two cultures of mathematics and biology): https://liorpachter.wordpress.com/2014/12/30/the-two-cultures-of-mathe… It was previously discussed here: https://news.ycombinator.com/item?id=8819811

It's like finance or law or any other field with some hothead cutthroat personalities in powerful positions. Where there is perceived power there is an attraction to people who feed off of power alone. Egos are a plenty and often cruel to others trying for the same goals. Lots of people are hypercompetitive rather than collaborative, and see any peer as an adversary or challenger rather than leaning into synergy. This kind of stuff would happen at conferences too, a couple notable cases of screaming matches from the audience to the presenter have happened. Sometimes vindictive professors will go to a rival professors's grad student's talk and just grill the poor kid, asking them impossible questions and being hypercritical. Usually the grad student's own professor has to step in and try to deflect the assholes line of questioning. Sometimes the professors are even in the same department. It's a total cringe moment for the rest of the audience, and is definitely a phenomenon that is slowly growing out of the field as professors retire. Take a look at the comment section of this blog to see it first hand: https://liorpachter.wordpress.com/2017/08/02/how-not-to-perform-a-diff…

It's tricky to pinpoint exactly when modern "bioinformatics" started. Some notable times are 1979 when the Los Alamos Sequence Database started, 1982 when that became GenBank, and certainly by 1990 when the Human Genome Project started. GenBank is "an open access, annotated collection of all publicly available nucleotide sequences and their protein translations" (https://en.wikipedia.org/wiki/GenBank ). Most journals now require that sequences be entered into GenBank before publication. I'll add a few historical observations. The early work at Los Alamos was on Sun machines running Sybase (1987, according to https://en.wikipedia.org/wiki/Sybase ). These are serious Unix fans, who quickly took to perl when it came out. I think some of the support for SybPerl came from bioinformatics; the SybPerl, OraPerl, and other *Perl systems helped create the extension system for perl5. (I am not able to verify that, though the author of SybPerl, Michael Peppler, consulted in the mid-1990s for "Research Genetics" http://www.peppler.org/resume.html .) Unix people in the early 1990s were serious perl fans. The most popular perl4 library for CGI programming was cgi-lib.pl, by Steven Brenner, a computational biologist/bioinformatics researcher. It was replaced in perl5 by CGI.pm, by Lincoln Stein, another bioinformatics researcher. This gives a hint that bioinformatics has not only a close connection to the technologies needed for the first dot-com era, but also that the field itself…

(Somewhat OT, but...) Constructive logic accepts the truth of every integer is either even or odd, but only because we have an effective means of determining which alternative holds for any integer. It does not accept the statement every real number is either rational or irrational; given a real number, say as a convergent series, we have no effective means of determining whether or not it is rational. I don't understand this. Consider the smallest number N so that any N points in the coordinate plane in general position contain a subset of 7 points that form the vertices of a convex polygon (taken from [0], recently discussed here on HN): is this number, which is definitely an integer, even or odd? We don't know, because we don't know what the number is. Isn't that the same situation we are in with regard to the convergent series? What's the difference? If we actually computed the decimal expansion of the value of the convergent series, we would be able to look at it and see whether it was rational or irrational. There is the minor difficulty that the expansion is in most cases going to be infinite, but so what? We're talking about mathematics here, not computation. [0] https://liorpachter.wordpress.com/2015/09/20/unsolved-problems-with-th… "Fifth grade" section

Teach biologists that? https://web.archive.org/web/20250115224532/http://www.dam.brown.edu/pe… https://golem.ph.utexas.edu/category/2014/12/can_one_explain_schemes_t… https://liorpachter.wordpress.com/2014/12/30/the-two-cultures-of-mathe… (2014) With someone actually trying in 2002, using TFB for inspiration & M2 as a very early Jupyter/sage https://web.archive.org/web/20250000000000*/https://mast.queensu.ca/~g… (These bloggers as well as creator of M1 may want to chip in a dozen cts)