Konstantin Korotkov 🇺🇦

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

Modeling of complexes between EsxA-EsxB and ESX-1 secretion-associated proteins EspB, EspC-EspA, EspF-EspE and EspJ-EspK from Mycobacterium #tuberculosis. #AlphaFold #ColabFold

[Roland Dunbrack 🏳️‍🌈 @rolanddunbrack.bsky.social @RolandDunbrack]

It appears the CCD code option is gone now from AlphaFold3 server.

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@jfernandobazan @RolandDunbrack @vfreis @joshim5 @chaidiscovery CCD Codes work on #AlphaFold3 server. Here is an MPro #SARS-CoV2 inhibitor from PDB 8u9h. The crystal structure in grey, #AF3 model colored by pLDDT.

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@AllThingsApx This figure is a bit misleading. The AF3 paper reports a runtime of 352s for the multi-GPU version of AF3, while the released single-GPU version of AF3 has a runtime of 62s for 1024 tokens.

[Yoshitaka Moriwaki @Ag_smith]

alphafold3_tools https://t.co/dqGCdmuaie
作りました。msatojsonはColabFoldのサーバーまたはcolabfold_searchで得られたa3mフォーマットのMSAファイルをいい感じにAlphaFold3のインプットJSONファイルに変換してくれるやつです

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@matsubara_m You can model myristic acid as a ligand and specify the covalent bond with the N atom of Gly.

[Gabriele Corso @GabriCorso]

Thrilled to announce Boltz-1, the first open-source and commercially available model to achieve AlphaFold3-level accuracy on biomolecular structure prediction! An exciting collaboration with @jeremyWohlwend, @pas_saro and an amazing team at MIT and Genesis Therapeutics. A thread!

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@chembioBryan For natural protein-protein interactions RF2-PPI should work better https://t.co/5pRXcxL53r. In your minibinders the interface is obvious, so knowing the target interface would be helpful. I guess/hope you targeted the active site of DltD 🙃 solving 6PFX structure was fun.

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@chembioBryan AF2/AF3 poorly discriminate binders vs. non-binders even for natural proteins using default MSA-generation, especially if there are multiple paralogs for the putative interacting protein pairs.

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@chembioBryan This result is not surprising. AF3 uses MSAs to derive co-evolutionary contacts to guide the predictions. For engineered binders - there is no co-evolution signal. Also, AF2/AF3 was not trained to discriminate binders vs. non-binders.

[Patrick Bryant @Patrick18287926]

Check out our (@FrankNoeBerlin) new paper where we improve the prediction of protein complexes by 'hacking' AlphaFold-multimer (AFM). Starting from an MSA, we use AFM to update what knowledge to sample from the MSA with gradient descent through the network. (1/2)

[Sergey Ovchinnikov @sokrypton]

Towards the end of the presentation I went down a bit of a rabbit hole trying to demonstrate that AF3 may still be learning to invert the convariance matrix, which is needed to extract the coevolution signal from input multiple sequence alignment (MSA) (1/9).

[Natalia Korotkova 🇺🇦 @nkorotk]

I am deeply honored to have been selected for this wonderful award. This research would not have been possible without the hard work of my wonderful trainees, lab members, and collaborators. @UKyMIMG #Streptococcus #IDRs

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@GntlmnScientist It's time to go back to the wet lab. Unlikely to be solved by MR even with a perfect model.

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@Matt_Arnold_Bio For #ColabFold-style pAE plots:
cmap = https://t.co/dRsjNCFRxo.get_cmap('bwr')

[Konstantin Korotkov 🇺🇦 @korotkov_lab]

@Matt_Arnold_Bio If you want to convert AlphaPickle to AlphaGreenPickle, i.e. generate pAE plots in standard #AlphaFold green-white colors, modify def plot_PAE in Colab notebook. Add
cmap = https://t.co/dRsjNCFRxo.get_cmap('Greens_r')
and edit
PAE = plt.imshow(self.PAE, cmap=cmap)
🙂

[Patty Champion @pattychampionND]

Do you want to be considered for an oral presentation at the Phage Meeting? Register and submit your abstract by June 10th for consideration for a talk! Molecular Genetics of Bacteria and Phages Meeting | Hosted By Wisconsin Union Conference Services https://t.co/vkAMEmsJM2

[Nina van Sorge @N_van_sorge]

More work to do….👇

[Jessi Blackburn @TheBlackburnLab]

Had an amazing time at the @TheVFoundation V Summit! Lots of exciting research from new V Scholars, serving on the mentoring panel was fun, and workshops were so helpful! Best part: catching up with @CarolineN_Smith, Director at the V Foundation (but still my lab superstar 🤗)

[Nina van Sorge @N_van_sorge]

The skin epidermis is populated with Langerhans cells. In this preprint, we provide initial proof of concept that LC may fuel the inflammatory cycle by connecting Staph aureus to T cell activation. Collaboration with @FemkevanWijk1 https://t.co/ymZzzKAnGq