Christopher Ing

Love "Noun Project" but... have any of these designers seen a molecule before?

Searching for a DOI in 2026

"Yeah, I’ve got a multi-agent stack doing end-to-end development. One handles planning and orchestration, another does test and verification, and a third just vibes and boosts morale."
"Oh, why am I needed? Taste, man."

I'm surprised, but not surprised, that ChatGPT citations often point to sites like ResearchGate. RG sucks but they seem to genuinely be helping piece together scientific literature "behind the paywall" that's enabling LLMs to deliver more value...

You have to be opportunistic with new technologies—mostly as a result of institutional paranoia. My surge in AI coding assistant use is primarily a race against the enshittification timeline.

Conventional wisdom: "Don't let your pipes freeze, be careful!!"
Refreshing advice: "Almost let your pipes freeze, your tap water will be ice cold and delicious!"

As far as I can tell, the real reason why antimicrobial peptides aren't in clinical development is that models are too good. Shame it's a solved problem =(

GitHub - haykgrigo3/TimeCapsuleLLM: A LLM trained only on data from certain time periods to reduce modern bias A LLM trained only on data from certain time periods to reduce modern bias - haykgrigo3/TimeCapsuleLLM

Okay, now try a protein LLM trained on only ancestral proteins to reduce modern bias (a simpler time without so many protein fitness influencers) github.com/haykgrigo3/T...

"Taken together, these results demonstrate that [method X] is a general-purpose model capable of aiding in all stages of designing and optimizing [Y] in real-world drug discovery campaigns." (...Well, except for ligand-receptor structure prediction)

Patsnap is really putting the "landscape" in IP landscape with this one. Looking forward to the minecraft server extension to properly explore the IP ecosystem (watch out for patent trolls...)

Never thought I'd get a chance to smell a supernova in my lifetime. Thankful.

Toronto property tax portal is like: you must enter this random 21-digit number! Then on another page their like, screw it, just put in your address we'll auto-populate it for you.

You'd be inclined to believe this innocent helper function from ChatGPT is not flawed/wrong ("atom_number" does not exist in biotite, and it's not safe to assume hydrogen atom names start with "H", i.e. if they were output from Rosetta).

2023: "de novo antibody design is solved" -AbSci
2024: "de novo antibody design is solved, we raised $1B on this" -Xaira
2025: "de novo antibody design is solved, the last time was jokes" -Chai, Nabla
20XX: "de novo antibody design is solved, can anyone hear me... anyone out there?" -Startup X

Improved targeted delivery of antisense oligonucleotide with an antibody mask Abstract. Antisense oligonucleotides (ASOs) are synthetic nucleic acid strands designed to modulate gene expression by binding to RNA transcripts. In brain

Trick-or-treaters weren't the only ones wearing a mask this Halloween, doi.org/10.1093/nar/...

Critics: "Boo, -latest/greatest AI drug discovery method- only works if you have a similar training example!"
Biotech Industry: "Hurry, let's find the smallest possible modification to this approved ligand which hits the identical target and epitope"

Real data tracking my head nodding at the Autechre tour stop in Toronto last night

“What fascinated me was the fact that you could potentially predict which molecule could have this effect using thermodynamic principles in silica” www.fiercebiotech.com/special-repo...

"We discovered a TfR1-targeted brain shuttle, check out the brain uptake" (Crook et al. 2020, doi.org/10.1016/j.jm...). "This radioligand does not efficiently cross the blood-brain barrier" (Lin et al. 2025, doi.org/10.1007/s002...)

When will a generative model pass the "Internet Turing Test"? That is, a test subject cannot distinguish an AI-generated internet from the real internet.

Scientists: answering the questions we were dying to know, but too afraid to ask

There's a form of "Personal Computer" Stockholm syndrome where an accumulation of weird tech issues makes computer use border on torture, with a surprising user tolerance threshold. Usually it's only identified by a 3rd party using your machine and remarking... "How the heck do you use this thing?"

"The Illiad" is kinda like "Three-Body Problem" in that their both about vengeful omnipresent beings fucking around with physics on earth.

Sequence logo's are so underutilized, even in protein design papers. Mainly because all the libraries to create them have zero imagination for applications. Represent ensembles of any residue-level property with symbol/color, or use pos/neg enrichment to show differential bias between any two sets!

Grant me the confidence of a scientist asking for help installing software in the Boltz2 webinar Zoom chat

Increasingly convinced there's collusion in "big spinach" to sell washed leaves "wet" in grocery store tubs to accelerate rot and increase sales 🤫🥬💧🤑

M1CR0B1AL1Z3R 2.0: an enhanced web server for comparative analysis of bacterial genomes at scale

leetspeak still alive in the bioinformatics community, academic.oup.com/nar/advance-...

Kinda like mining bitcoin

Love and respect the four-digit PDB code convention, and will miss it. I like how you can use it to roughly "date" structures. Like, 3J5P as a marker for the start of the cryo-EM membrane protein era in 2013.

As if the big scientific publishers weren't bad enough, they sometimes just take journals offline. "Current Opinion in Drug Discovery & Development" ran from 1998-2010. Owned by Thomson but almost all links are broken with no other hosts (that I can tell), pubmed.ncbi.nlm.nih.gov?term=%22Curr...