Thursday, November 30, 2023

An AI helped make this month's JPR cover!


Another bad week for the blog, but I've been under multiple deadlines and I'm sort of tired of writing. But here we are at 3am typing in the orange box instead of working on things that are due in like 6 hours. 

This probably isn't the first JPR cover generated with assistance from an AI? I don't know, but this one used help from the DreamStudio AI. 1,000 iterations cost $10 and this one didn't even use 100 of those. will generate some whacko nonsense if you don't use the right terms. And if you give it text it does not know what to do. Is that kanji? Or is it just random lines? Have it generate pictures, fine tune them and then add everything else. Is that yogi the bear doing yoga? That's funny, but not at all what I was trying to generate. 

One of my absolute favorite parts of this study is this candid statement in all capital letters. Why does a static gun need to be shaped in any way like an actual gun? As a lot of people know, I was very physically close to one of the most horrific mass shootings in US history.  It didn't do great things for my brain - which, to be fair, was sort of mediocre to begin with. Imma patent a static gun that looks like a pencil or one of those penguin like animals they have in Maine or something soon, cause working at 3am in a quiet inner city building holding something that looks like a f'ing gun deserves notice as a stupid thing. 

Oh yeah, this is the paper! Yet another single author paper? Weird. Okay, so there was a student who was supposed to do this study as their final 1-1.5 years for their PhD, but then that student just never came back to lab. So I had spent all this money on cell lines and reagents and hundreds of hours on training and planning a study I was emotionally invested in and I was a little annoyed about the situation. Instead of being all angry or sad or regretful for hours of my life when I don't have an infinite number of hours left, I cut some experiments from the study and knocked out the majority of the planned work over the Xmas holiday break by myself. 

Honestly, the coolest part of the study is the global metabolomics. Also - this might be the first published paper to use the SimpliFi Cloud based multi-omics data interpretation system? I've been a beta tester of this ProtiFi project since the beginning and it's super cool for sharing processed and interpreted data. For some reason I don't remember, I broke this study into separate SimpliFi projects. Here is one of them.  I hope we'll have something to directly import global metabolomics data from Compound Discoverer into SimpliFi. About got that node configured, it's just an abacus issue. 

We just uploaded some new data to MASSIVE that is way better than this, btw. The people doing the prep now are better at sample prep than I am. 

No comments:

Post a Comment