Stephen MacKinnon

@BoWang87 I’d say ‘conditionality’ is the biggest differentiating factor for biological data. Biological systems have very complex interdependencies, moreso than anything I can think of in language or other domains.

The partnership between Nvidia and leading biotech and pharmaceutical companies brings together cutting-edge computing with vast biological knowledge and data. Recognizing the unique challenges that differentiate the fields of biology from languages, I am wondering: What else do we need to create another AlphaFold moment or a chat-GPT moment in biology? Thoughts?

Have a quick 👀 @ss_mackinnon's newest post, where he discusses how #SAMR is similar to the role that @DeepMind 's #AlphaFold plays in Drug Discovery. 👉blog.cyclicarx.com/recognizing-tr…

@ddegraaf @PearlF @mmameg @Cyclica @NaheedKurji AF2 models do provide excellent protein shape/topology, but specific atomistic detail may be lacking. For instance, a drug binding model trained on general pocket features would then be more appropriate than one that trains on specific specific atomic geometry.

@ddegraaf @PearlF @mmameg @Cyclica @NaheedKurji Oh, the structures are really good, but not perfect. As long as the application/ inference factors in the model’s caveats, the scale of structures can provide significant value.

@sokrypton Oh good question! I think of Structural Bioinformatics as extracting biological insight from looking at many structures rather than a single one. AF/RF makes it tricky but not impossible.

AlphaFold/RosettaFold users: I was asked to give a talk summarizing how the community has been using AF/RF for "structural bioinformatics". I'll focus on deep structural homology/similarity insights that would otherwise not be possible with just sequences. Please share! (1/2)

@NaheedKurji Lol that photo

From part time scientist to Chief Platform Officer, Stephen Mackinnon has been one of Cyclica’s scientific strongholds for 8 years. He’s a close friend, a mentor, a brilliant scientist, a better leader and someone I’ve leaned on since the very beginning. To many more years.

Thanks to Dene Moore & @globeandmail for featuring us, @windemut, @ss_mackinnon, & our recent collaborator Cleber Trujillo from @VyantBio. #drugdiscovery #artificialintelligence #MachineLearning

As part of our AlphaFold series, our third post explores the breakdown of proteome coverage by protein class authored by @ss_mackinnon & David Kuter. Link to the post: lnkd.in/eNpdnv9y #AlphaFold #artificialintelliegence #drugdiscovery #MachineLearning

@AndrewYNg Using data with quantifiable uncertainty… ie, mixing high confidence and low confidence data together effectively

I would love your thoughts on this: Data-centric AI is still an emerging field, but what do you think are the key pillars of data-centric AI? E.g., if you were reading a textbook on this nascent field, what are some major topics you’d like the book to include?

With Mitacs support, @Cyclica has partnered with @BoWang87 of @UofT to examine the viability of existing drugs as treatment for #COVID19. They will apply #AI to mine biological networks formed between virus & human proteins & predict drugs that can interrupt these interactions.

Thanks for the kind support from @MitacsCanada and partnership from @Cyclica. We aim to utilize #AI to expedite the drug repurposing discovery for #COVID19. Please DM if you are passionate about #AI and #drugdiscovery. Position available @VectorInst @PMunkCardiacCtr @UHN_Research

ATAI Life Sciences and Cyclica launch Entheogenix Biosciences Inc., a JV that will revolutionize drug development for mental health disorders by designing new compounds inspired by desirable psychedelics properties. businesswire.com/news/home/2019… #mentalhealth #medicines #depression

This could be the future of shrooms and other psychedelics. bit.ly/2WS7vvR

After a long deliberation the judges arrived at a unanimous winner of our $250k startup pitch competition. Congrats to Paradox Immunotherapeutics!!

@atomoton But wouldn't that only happen twice in your life? I would argue that it is more convenient than a continuous growth pattern.

@LeelaDodda @iwatobipen Take a look at selfies instead of smiles. Kinda similar in concept, but has a much higher rate of valid molecules. github.com/aspuru-guzik-g…

Graph based generative model of molecule #DGL #RDKit #chemoinformatics iwatobipen.wordpress.com/2019/09/10/gra…

new: "Machine learning models for hydrogen bond donor and acceptor strengths using large and diverse training data generated by first-principles interaction free energies" jcheminf.biomedcentral.com/articles/10.11…

@BrookfieldIIE @CIFAR_News An important initiative for Canada!

Learn more about our cross-country series of labs we designed + delivered in partnership with @CIFAR_News to explore the public policy implications of #AI brookfieldinstitute.ca/report/rebooti…

@GoogleDevExpert @Cyclica Steve Constable presenting our Ligand Design platform built on the Google Cloud Engine at the NA GDE Summit, @mad_sha @Cyclica @GoogleDevExpert

Steve Constable discussing Cyclica - a company that leverages AI & biophysics to remove the mystery and get drugs to the market faster at a lower cost @Cyclica #NAGDESummit2019