@npokorzynski

Asst Prof @OregonState Microbiology | fmr Postdoc @YaleMicroPath | Bacterial metabolism, virulence, & abx susceptibility | he/him | Views = mine @PokorzynskiLab

 Corvallis, OR
pokorzynski-lab.squarespace.com
Joined March 2016
  • Tweets3.5K
  • Following1.9K
  • Followers811
182Media
Very happy to share the second paper from my postdoc, published today @PLOSBiology. Here, we report that the master regulator of Salmonella virulence usurps the reign of cAMP over metabolism under infection-relevant conditions. journals.plos.org/plosbiolog…

The master virulence regulator PhoP dictates carbon metabolism by controlling cyclic AMP synthesis...

Inside macrophages, Salmonella faces magnesium scarcity that disrupts cAMP synthesis and carbon metabolism. This study shows that the PhoP regulator restores metabolic control by promoting Mg2+...

journals.plos.org
2
2
8
780
I'm pleased to share my contribution, alongside my former advisor, Eduardo Groisman, to @AnnualReviews of Microbiology volume 80: Mg2+, ATP, and the Control of Bacterial Metabolism.
1
1
5
311
more replies
Here, we discuss how many of these metabolic changes arise because ATP is a precursor of nucleotide second messengers which control bacterial metabolism.
1
26
The control of metabolism by ATP-derived signaling molecules rather than ATP itself may help explain why the cellular ATP concentration greatly exceeds that needed to operate ATP-dependent enzymes.
22
You know someone was asleep at the wheel when you see a reputable journal publish a paper describing the “trichloroacetic acid cycle” in the abstract…
1
247
A brief story on why this is wrong-headed and doomed to fail: My lab is currently studying the regulation of a gene. We think the gene is doing a surprising amount of stuff, and it has some regulatory features that we are interested in from prior work. 1/11
Jason Kelly
@jrkellyOct 10
If we want to unlock biological engineering we need to kill the manual lab bench. I've tried almost everything else over the last 20 years😛and PhDs still having to be in the lab is the rate limiter.
9
5
61
12,957
more replies
Automation views humans as the problem in scientific discovery when they are in fact the *only* generative element. When you take humans out of the experimental equation, you will lose the generative potential that fuels discovery. 10/11
1
4
27
1,316
Now, if you want to screen a million drugs against cultured cells, be my guest and automate it. However, you and I have very different definitions of “science,” in that case. 11/11
2
27
1,262
I suspect entire grant reviews I’ve received have been AI generated - which is annoying because AI critiques are pedantic but much more egregious is the ethical breach of feeding my ideas/data to an agent without my consent. Perhaps now we’re ready to have that conversation.
Joel Fish
@joelwatsonfishOct 10
Lots of folks are dunking on Tristan here, but towards the end of the clip he points out a serious security issue. When you submit a grant proposal to NSF, you basically have to outline the research that you will do in the next few years with enough details that an expert in the field can assess whether it’s likely that you’ll complete it. All panelists in that group of reviewers have access to that proposal, and confidentiality is assumed and usually isn’t a problem. But if any panelist pops the proposal into an LLM (say, to summarize or assess), confidentiality has been broken and a private research plan has just been given to a for-profit competitor. This is an accidental breach, but a serious breach still. Can’t OAI/ANT just scrape these inputs for ideas to try?
1
5
194
I think this is true and has not been meaningfully addressed by the community. Agents provide feedback that is granular in a way humans do not, and that granularity is often off the mark by orders of magnitude.
Andrew S. Rosen@Andrew_S_RosenSep 26
Replying to @anshulkundaje
Worse, I don't think it will be more random. I think it will be that the proposals reviewed by LLM will be ranked lower than those reviewed without. The LLM reviews are always critical, whether superficially or not, even for exceptional work. This should not be done.
1
2
592
For example: a criticism over a mutation “readily acquiring suppressors” not because it is true (and any domain expert would know this) but because *suppressors of the mutation are well characterized.*
1
70
In just a year on the job, I’ve received fully AI reviews that provided such probing critiques as “the lab is too new” for a “new investigator award.” It goes without saying that this sets a really bleak standard, putting aside the obvious ethical concern of feeding grants to AI.
2
31
Load more