Wow! That’s rather embarrassing not to be there for that ( sorry about that- I’m mid-air -almost in Perth for a talk at the protein meeting tomorrow) and rather humbling as well. 🙏

It was a great meeting! Highly recommended. Sorry to have missed the last and best part- all those amazing early career speakers and poster prizes!!! Congratulations everyone. Wonderful job to the organisers.

Follow Tom here: @tlitfin.bsky.social

And preprint: www.biorxiv.org/content/10.1...

I haven’t tried it. my day-to-day doesn’t use those tools that often. I can give it a go and let you know. I’m sure there will be uses that are helpful. I am very grateful for Alphafold, Boltz, RF diffusion, and BindCraft. They certainly have changed my day to day. There are other bottle necks now.

I’ve seen some doozy traps that people have fallen into that aren’t structural biologists. Could we train another model to analyse the output better? Probably. It still isn’t at the ‘walk away’ phase.

You can give biomolecular DL programs a protein sequence to fold and they will output a result. You need to really understand all the metrics to know if that output is any good. There certainly are examples with high confidence metrics that are rubbish. You need to really be an expert to judge.

I know. It’s sad. Some people in their desire to control and dominate make me sad. There is no prize at the end of life and you can’t take it all with you. Death is inevitable.

One thing that they have taught people with this grab for power is the value of data. Make no mistake that some people have noticed this and things are changing about how data is shared. I can’t imagine what 5 years down the road will look right now. The game has changed. Adapt or die.

We don’t have the fundamental data required to train an AI to do these tasks yet. Science in this area is extremely difficult and there are many examples of very smart people changing fields to ‘fix’ it and who have disappeared into obscurity.

For instance the claim it will solve cancer is deluded. The people that say that don’t have a clue what cancer is. Alphafold3 can’t even tell you which metal atom is in a protein structure. It certainly is not ‘designing’ drugs right now. This is all about attracting venture capital and isn’t truth.

They underestimate the complexity of biology and over estimate human understanding. There is a skill shortage required to scale most of the AI assisted results. Deep understanding of the limitations is required. For now the weight is still on the scientist’s foot. They may have gone hard too early

Assuming deep sea organisms won’t last long and so this is the best thing we can do to try to ‘observe them’. Certainly can’t freeze thaw them and prepare later unfortunately. Heroic taking a plunger onto the seas!

At the time the cultures were so complex it was impossible to tell precisely which organism had the vesicles. The cultures contain many different types of archaea- a real treasure trove, and lots of bacteria with lots of ‘weirdness’.

Yes they are! Might have been an oversight, but these cultures were established at UNSW in 2019 in @brendanburns999.bsky.social ‘s lab and funded by Australian research grants. The first tomograms that revealed the exciting internal structures were carried out by @debnathghosal.bsky.social

I was such a huge Frenzal groupie…😂