An intriguing answer engine for biology

👇

My colleague Germán Plata pointed me to a super interesting "chatGPT" for biology, bioloGPT (link), developed by Conner Lambden. It may sound boring -you know, yet another chatbot, or why not use Perplexity.

➡ But there are a couple of features that make it stand apart. Here is an example: I asked the question: "What is LL-37?". It came back with a reasonable answer with sources and references, not that different from others -I can't believe we are now used to this degree of sophistication with "answer engines":

"LL-37, also known as hCAP-18, is a human cathelicidin antimicrobial peptide that plays a crucial role in innate immunity. It is expressed in various tissues and cell types, contributing to the body's defense against infections."

💡 It also came up with a "Key Insight", which I found to be spot on:

"LL-37 is a multifunctional peptide that not only exhibits antimicrobial properties but also plays a significant role in modulating immune responses, highlighting its importance in both innate immunity and potential therapeutic applications."

💰 But perhaps the biggest differentiator is a set of 3 short sections (screenshot below):

• Potential Hypothesis (to be tested)

• Hypothesis Graveyard (things that have been disproved)

• Potential Experiments (to test hypotheses, though not necessarily the same ones as in 1.)

I am assuming it is just the beginning. It has enormous potential, especially if it can start generating entirely new hypotheses, which does not seem to be the case at this point based on cursory use of the tool. Go Conner Lambden!