Entering edit mode
10 months ago
Sasha
▴
840
Hey everyone,
Today we are publically releasing a new large language model tool for brainstorming scripts based on the types of samples you have access to.
The main use case that we want to solve with this UI is the cold start problem when you're staring at a blank screen trying to figure out what to run next. For each pipeline, before it is generated, you can also edit the software stack used so that you can incorporate your feedback before the script is generated.
Try it out here: https://ai.tinybio.cloud/experiment-generator
sasha -
as you know, science tends operate in multi-year cycles, ideally with meticulous planning for exactly what will be done based on a careful understanding of a biological problem.
more to the point, grant funding is at present esp. scarce, so, if one has spent a several grand per sample on animal storage, handling, sequencing, then analysis and data storage, etc. for 12 samples, i sincerely hope they know what they are going to do with it.
for this reason, i was suprised you didnt say the main use-case is as a pedagogical tool
Absolutely - definitely a great tool for learning and teaching as well.
During our interview process, we heard that folks wanted to know the set of different other analyses & other software pipeline combinations that could be run on samples that they had collected. To your point, they did have the research question in mind when they started - but this tool is meant to give other suggestions as well so that it's easier to get that next grant and start that next pipeline.
yes thats true. as a way to get extra mileage out of the data - could be useful.
could be helpful to build in functionality that looks at the output from analyses already run to suggest problems or pitfalls that could be limiting the analyses completed as well.