Science The silly journal club

[Skorpio]

Post articles that are a bit silly and a bit cool here.

This article below is exactly what you think it is. Scientists trained monkeys to play pac man by rewarding them with various drops of fruit juice based on different actions (some for eating small dots, more for eating the power ups, even more for eating ghosts and fruit).

Afterwards they used machine learning to identify the strategies the monkeys were using and it was found that like humans they use multiple different strategies, switching depending on context.

Monkey plays Pac-Man with compositional strategies and hierarchical decision-making

 

[izo]

So cool, Sigma aldrich started this as April fool and some chemists actually delivered…

 

[izo]

Also this ridiculous paper:

2012 - A Simple and Convenient Synthesis of Pseudoephedrine From Methamphetamine.pdf

drive.google.com

 

[Jabberwocky]

andre geim, famous for making graphene, and michael berry, head of the physics department at what i think americans would call my Alma Mater (actually have links to both the institutions involved in this v important research lol), on levitating frogs:

ShieldSquare Error

iopscience.iop.org

this work ensured that Geim received an IgNobel prize, to go with his boring Nobel.

popularised account here:

In 1997, Scientists Made A Frog Levitate

In 1997, Scientists Made A Frog Levitate

www.iflscience.com

 

[Skorpio]

Also this ridiculous paper:

2012 - A Simple and Convenient Synthesis of Pseudoephedrine From Methamphetamine.pdf

drive.google.com

That one is such a classic!

These 2 paragraphs are my favorite
"
A quick search of several neighborhoods of the United States
revealed that while pseudoephedrine is difficult to obtain,
N-methylamphetamine can be procured at almost any time
on short notice and in quantities sufficient for synthesis of
useful amounts of the desired material. Moreover, according
to government maintained statistics, N-methylmphetamine
is becoming an increasingly attractive starting material for
pseudoephedrine, as the availability of N-methylmphetamine
has remained high while prices have dropped and purity
has increased2
. We present here a convenient series of
transformations using reagents which can be found in most
well stocked organic chemistry laboratories to produce
psuedoephedrine from N-methylamphetamine.
While N-methylamphetamine itself is a powerful
decongestant, it is less desirable in a medical setting because
of its severe side effects and addictive properties3
. Such
side effects may include insomnia, agitation, irritability,
dry mouth, sweating, and heart palpitations. Other side
effects may include violent urges or, similarly, the urge to be
successful in business or finance."

 

[izo]

Other side
effects may include violent urges or, similarly, the urge to be
successful in business or finance."

Lol

 

[izo]

Another silly paper:

 

[Jabberwocky]

i haven't actually read this paper, but it deserves a mention cos i am a teenage boy stuck in a mid-30s woman's body:

Vagococcus penaei sp. nov., isolated from spoilage microbiota of cooked shrimp (Penaeus vannamei) - PubMed

A polyphasic taxonomic study, using phenotypic, phylogenetic and genotypic characterization, was performed on five Gram-stain-positive, catalase-negative, coccus-shaped Vagococcus-like bacteria isolated from the spoilage microbiota of cooked shrimp. Comparative 16S rRNA gene sequence analysis...

pubmed.ncbi.nlm.nih.gov

imagine you have a new bacterium of genus vagococcus, and you decide to call it penaei. i get that its because it was found in a shrimp, or penaeus, but still they could have found a different distinguishing feature to name it for.

 

[Skorpio]

i haven't actually read this paper, but it deserves a mention cos i am a teenage boy stuck in a mid-30s woman's body:

Vagococcus penaei sp. nov., isolated from spoilage microbiota of cooked shrimp (Penaeus vannamei) - PubMed

A polyphasic taxonomic study, using phenotypic, phylogenetic and genotypic characterization, was performed on five Gram-stain-positive, catalase-negative, coccus-shaped Vagococcus-like bacteria isolated from the spoilage microbiota of cooked shrimp. Comparative 16S rRNA gene sequence analysis...

pubmed.ncbi.nlm.nih.gov

imagine you have a new bacterium of genus vagococcus, and you decide to call it penaei. i get that its because it was found in a shrimp, or penaeus, but still they could have found a different distinguishing feature to name it for.

I feel like it's somewhat burying the lede to mention that peneaus is the genus of prawns. My whole day is going to be so much less mature!

 

[Jabberwocky]

I feel like it's somewhat burying the lede to mention that peneaus is the genus of prawns. My whole day is going to be so much less mature!

haha we all need to be puerile sometimes. does make you wonder about the anatomy of the person that named them....

 

[izo]

Another funny paper:

Molecular Sonification for Molecule to Music Information Transfer

Organic chemical structures encode information about a molecule’s atom and bond arrangement. The most established way to encode a molecular structure is through line drawing, although other representations based on graphs, strings, one-hot encoded labels, or fingerprint arrays are critical to...

chemrxiv.org

 

[izo]

Another good idea: turning wine into blood:

 

[thegreenhand]

the dead salmon fMRI study is a classic

http://f20.blog.uni-heidelberg.de/files/2021/07/bennett-etal-2009-dead-salmon.pdf

the researchers found 'statistically significant' activation in the brain of this dead salmon after showing it various pictures intended to invoke emotions.

it was more so a demonstration of the need for a different way of doing statistics on fMRI data to avoid false postives, but funny nonetheless

 

[Jabberwocky]

the researchers found 'statistically significant' activation in the brain of this dead salmon after showing it various pictures intended to invoke emotions.

it was more so a demonstration of the need for a different way of doing statistics on fMRI data to avoid false postives, but funny nonetheless

awesome!!! i love things like that.

statistical significance is a slippery concept. we do need some cut off to determine what results we care about and what doesn't count as statistically significant but it is so easy to cherry pick results or look at the wrong combination of variables to give you something you want. its easy to spot if you're looking for it, but so very compelling if you don't. for shit my professional reputation rides on if the raw data aren't available then i don't trust it.

 

[thegreenhand]

awesome!!! i love things like that.

statistical significance is a slippery concept. we do need some cut off to determine what results we care about and what doesn't count as statistically significant but it is so easy to cherry pick results or look at the wrong combination of variables to give you something you want. its easy to spot if you're looking for it, but so very compelling if you don't. for shit my professional reputation rides on if the raw data aren't available then i don't trust it.

It definitely is a particularly finicky concept

what always struck me as interesting is how much individual disciplines vary in ‘acceptable’ p-value. for example, physicists seem to take p = .001 as alright, while researchers in psychology might see p = .05 as passing

tbh i dont think i really understand the full definition of a p-value lol

but yeah imo all scientific literature ought to have completely open data. i really don’t think the peer review system can work properly without it tbh

 

[Conscious-signal-447]

Please allow me to present my rather pragmatic contribution to this thread:

Frontiers | The Psychonauts’ World of Cognitive Enhancers

BackgroundThere is growing availability of novel psychoactive substances (NPS), including cognitive enhancers (CEs) which can be used in the treatment of cer...

www.frontiersin.org

“The psychonauts world of cognitive enhancers”. A handy little list of psychoactive substances identified by AI “web crawling” technology.

*bows and gracefully withdraws*

 

[Jabberwocky]

tbh i dont think i really understand the full definition of a p-value lol

can i take this as a cue to point you towards this paper, which says they are a bit silly and therefore don't warrant time being spent understanding:

https://www.biorxiv.org/content/biorxiv/early/2019/04/06/377978.full.pdf

they have this web client for making plots:

Estimation Stats

www.estimationstats.com

even within physics, your confidence interval varies massively. particle physicists could have errors that meant getting their results within an order of magnitude of the expected value was considered OK. quantum mechanics routinely predicts things correctly to something like 15 sig fig iirc. whereas when i was doing pre clinical research they just always used 0.05, but then they always used exactly the same statistical tests even when the assumptions required for those tests to be valid were not applicable, so that was wild.

“The psychonauts world of cognitive enhancers”. A handy little list of psychoactive substances identified by AI “web crawling” technology.

*bows and gracefully withdraws*

excellent!! thank you.

next they should build an AI to predict the effect of each of the substances so i don't have to try em all.

 

[thegreenhand]

can i take this as a cue to point you towards this paper, which says they are a bit silly and therefore don't warrant time being spent understanding:

https://www.biorxiv.org/content/biorxiv/early/2019/04/06/377978.full.pdf
they have this web client for making plots:

that was a very insightful paper. and I especially like that they created accessible tools to help people make changes

even within physics, your confidence interval varies massively. particle physicists could have errors that meant getting their results within an order of magnitude of the expected value was considered OK. quantum mechanics routinely predicts things correctly to something like 15 sig fig iirc. whereas when i was doing pre clinical research they just always used 0.05, but then they always used exactly the same statistical tests even when the assumptions required for those tests to be valid were not applicable, so that was wild.

Do you think that was because biologists tend to lack the math/stats knowledge to properly use these sorts of tests or is it some other reason?

 

[Jabberwocky]

that was a very insightful paper. and I especially like that they created accessible tools to help people make changes

glad you enjoyed it!! i'm a big fan, anyone mentions anything remotely relevant and i point them to it.

Do you think that was because biologists tend to lack the math/stats knowledge to properly use these sorts of tests or is it some other reason?

its not all biologists, well if you consider medics biologists. cos we had a med student taking a year out to get a masters by research, and she was bothered too. some of it is inertial, cos a lot of papers do it so if you don't do what the peer reviewer is familiar with then even if its the right thing to do technically, you might get push back.

but mostly its not lack of knowledge (though i doubt my colleagues did have detailed knowledge), its totally not caring. their main job is designing and performing experiments that take months, and then you have to do these analyses for an afternoon when you get some results. so they never do it often enough to get good at it, and if your interest is more in the experimental side then its less of a concern. plus the predictions are almost always qualitative, at least in what i was doing, so even when you do get to do data analysis, there is never really scope for detailed numerical analysis. like say you just care about whether there is 'statistically significant' up/down regulation of a particular set of genes, you use a t test to determine significance, but your sample size is tiny, data are not numerically distributed, and i never personally calculated a variance function so fuck knows whether that had the required form. but to publish a paper in a field where that's what it used, but with a more appropriate test, would probably result in reviewers/editors telling you to do it the 'correct' way.