Please check our latest paper on using @deep-mind.bsky.social AlphaFold 3 to study NLR immune receptors. Here we use AI-guided structural modelling to distinguish between sensor and helper NLRs! 👇🧵 (1/10) #EvoMPMI #NLRbiology #PlantPath #PlantSci doi.org/10.1101/2024... - ThreadSky

yusugihara.bsky.social • 87 days ago

Please check our latest paper on using @deep-mind.bsky.social AlphaFold 3 to study NLR immune receptors. Here we use AI-guided structural modelling to distinguish between sensor and helper NLRs! 👇🧵 (1/10) #EvoMPMI #NLRbiology #PlantPath #PlantSci
https://doi.org/10.1101/2024.11.24.625045

Comments

alxfed.bsky.social•87 days ago

Just so that you would know, this is a fake account of DeepMind that you linked to.

yusugihara.bsky.social•87 days ago

NLR immune receptors can be functionally organized into genetically linked pairs. The monocot paired NLRs, such as Pikm, Pia, and Pii, have evolved from distinct phylogenetic clades to function together as biochemically specialized sensors and helpers. (2/10)

yusugihara.bsky.social•87 days ago

Sensor NLRs often have unconventional integrated domains (IDs), which helper NLRs lack. But is there another way to distinguish paired NLRs? And can we differentiate sensor vs. helper NLRs in silico when neither one has an ID annotation? (3/10)

yusugihara.bsky.social•87 days ago

We recently showed that Nobel Prize-winning AlphaFold 3 (AF3) can be used to model NLR oligomers with high confidence. In this study, we wondered whether paired NLRs would produce distinct AF3 predictions. (4/10)
https://x.com/amiralito_/status/1803755420177039792

yusugihara.bsky.social•87 days ago

We started with AF3 modelling of 8 known NLR pairs, analyzing their oligomeric states. Interestingly, helper NLRs consistently showed higher confidence scores (pTM) than their sensors. (5/10)

yusugihara.bsky.social•87 days ago

Using the 50 oleic acids trick as a membrane proxy, helper NLRs consistently formed funnel-shaped structures contrasting with sensor NLRs. This indicates that AF3 can distinguish helpers from sensors. (6/10)

yusugihara.bsky.social•87 days ago

Next, we applied AF3 to 10 rice NLR pairs described in a previous study. All five pairs with an integrated domain (ID) annotation produced higher pTM scores for the helper NLR, further supporting AF3's ability to distinguish between sensor and helper NLRs. (7/10)

lambrosf.bsky.social•87 days ago

Excellent work, congratulations! Do you think that this approach can be used at scale? Also, does this work for finding matching effector and R gene pairs?

yusugihara.bsky.social•87 days ago

Thanks! Since the AF3 source code is now available, this can be extended at scale, definitely. But I'm not sure for the R-AVR pair.

Comments

Posting Rules

Reply