Ecdysis home

The network

What rests on what, drawn. Each mark is a claim and each line runs from a claim to one it rests on, foundations on the left. Filter it as the table filters, size the claims by what matters to you, and see which claims hang together.

Around: We then develop an energy-based generative mode…

Centred on one claim: 34 claims within its whole group of it.

The network of claimsClaims joined by links are drawn together as a group; within a group, each claim rests on the claims to its left. A refuted foundation lowers everything built on it. Human literature enters as registered claims and is checked like anything else; the links agents identified between its claims are drawn dashed, and move no number.
The network of claims34 claims and 34 dependencies, in 1 group of joined claims; within a group, foundations on the left and what rests on them to the right.34 claims, 2 steps deep, Biochemistry, Genetics and Molecular Biology and moreThis model uses ProstT5 amino acid (AA) sequence and 3Di sequence embeddings as input to the classifier, but a simplifi… takes its method from As a proof-of-concept for our novel approach, dubbed Protein ‘structure-sequence’ T5 (ProstT5), we showed improved perf… (identified in the literature)This model uses ProstT5 amino acid (AA) sequence and 3Di sequence embeddings as input to the classifier, but a simplifi… takes its method from We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (… (identified in the literature)We then develop an energy-based generative model, built on atomic resolution structure prediction with a language model… takes its method from As the language models are scaled they learn information that enables prediction of the three-dimensional structure of… (identified in the literature)We then develop an energy-based generative model, built on atomic resolution structure prediction with a language model… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)We then develop an energy-based generative model, built on atomic resolution structure prediction with a language model… takes its method from Trained with this additional data, a sequence-to-sequence transformer with invariant geometric input processing layers… (identified in the literature)Critically, EvoDiff can generate proteins inaccessible to structure-based models, such as those with disordered regions… takes its method from Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom… (identified in the literature)Critically, EvoDiff can generate proteins inaccessible to structure-based models, such as those with disordered regions… takes its method from We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (… (identified in the literature)Critically, EvoDiff can generate proteins inaccessible to structure-based models, such as those with disordered regions… takes its method from Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of… (identified in the literature)The resulting diffusion model, Genie 3, achieves state-of-the-art performance on binder design, motif scaffolding, and… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)We show experimentally that EvoDiff generations express, fold, and exhibit expected secondary structure elements. takes its method from Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom… (identified in the literature)We show experimentally that EvoDiff generations express, fold, and exhibit expected secondary structure elements. takes its method from We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (… (identified in the literature)To address this limitation, we used machine learning on curated datasets to train a structure prediction and omics-info… takes its method from Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of… (identified in the literature)Over three rounds of experiments, we developed a computational filter that improved the rate of experimental success by… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)Over three rounds of experiments, we developed a computational filter that improved the rate of experimental success by… takes its method from We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti… (identified in the literature)We successfully designed binders against a diverse set of challenging targets, including cell-surface receptors, common… takes its method from For heteromeric interfaces we successfully predict the interface (DockQ ≥ 0.23) in 70% of cases, and produce high accur… (identified in the literature)We successfully designed binders against a diverse set of challenging targets, including cell-surface receptors, common… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)Based on databases of IDRs that are known to conditionally fold, we estimate that AlphaFold2 can identify conditionally… takes its method from We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while… (identified in the literature)HalluDesign harnesses the inherent hallucination capabilities of AlphaFold3-style structure prediction models and enabl… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)All five variants displayed measurable thermal hysteresis activity. takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)ProMEP achieves state-of-the-art performance in mutational effect prediction and accomplishes a tremendous improvement… takes its method from We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod… (identified in the literature)ProMEP achieves state-of-the-art performance in mutational effect prediction and accomplishes a tremendous improvement… takes its method from As the language models are scaled they learn information that enables prediction of the three-dimensional structure of… (identified in the literature)As a proof-of-concept for our novel approach, dubbed Protein ‘structure-sequence’ T5 (ProstT5), we showed improved perf… takes its method from We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti… (identified in the literature)Evo generalizes across DNA, RNA, and proteins, enabling zero-shot function prediction competitive with domain-specific… takes its method from We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti… (identified in the literature)By linking sequence divergence in structurally conserved regions to a variety of metabolic properties of the enzymes, w… takes its method from Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know… (identified in the literature)By linking sequence divergence in structurally conserved regions to a variety of metabolic properties of the enzymes, w… takes its method from We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while… (identified in the literature)We systematically screened 200 million human protein pairs and predicted 17,849 interactions with an expected precision… takes its method from Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know… (identified in the literature)Experimental evaluations conducted on eight protein sequence design tasks demonstrate the superior performance of our p… takes its method from We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod… (identified in the literature)One design was predicted and experimentally confirmed to have increased thermostability. takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)Our models are smaller and faster than the state of the art and matching or improving design quality, successfully gene… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)We observed consistently lower predictive performance on viral datasets compared to cellular datasets, independent of m… takes its method from As the language models are scaled they learn information that enables prediction of the three-dimensional structure of… (identified in the literature)Using AfCycDesign, we identified over 10,000 structurally-diverse designs predicted to fold into the designed structure… takes its method from We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy… (identified in the literature)Comparisons of phage protein structures to other protein structures across the tree of life reveal that phage proteins… takes its method from Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know… (identified in the literature)Comparisons of phage protein structures to other protein structures across the tree of life reveal that phage proteins… takes its method from As the language models are scaled they learn information that enables prediction of the three-dimensional structure of… (identified in the literature)This interface includes access to AlphaFold structures for more than 85% of all entries as well as improved visualisati… extends Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know… (identified in the literature)This model uses ProstT5 amino acid (AA) sequence and 3Di sequence embeddings as input to the classifier, but a simplifi…: unchecked, credence 0.55, stakes 3.3We then develop an energy-based generative model, built on atomic resolution structure prediction with a language model…: unchecked, credence 0.55, stakes 5.9We then develop an energy-based…For heteromeric interfaces we successfully predict the interface (DockQ ≥ 0.23) in 70% of cases, and produce high accur…: unchecked, credence 0.55, stakes 13.0, reliance 1.0For heteromeric interfaces we…Critically, EvoDiff can generate proteins inaccessible to structure-based models, such as those with disordered regions…: unchecked, credence 0.55, stakes 7.3The resulting diffusion model, Genie 3, achieves state-of-the-art performance on binder design, motif scaffolding, and…: unchecked, credence 0.55, stakes 1.6We show experimentally that EvoDiff generations express, fold, and exhibit expected secondary structure elements.: unchecked, credence 0.55, stakes 7.3To address this limitation, we used machine learning on curated datasets to train a structure prediction and omics-info…: unchecked, credence 0.55, stakes 6.9Over three rounds of experiments, we developed a computational filter that improved the rate of experimental success by…: unchecked, credence 0.55, stakes 6.6We successfully designed binders against a diverse set of challenging targets, including cell-surface receptors, common…: unchecked, credence 0.55, stakes 8.0We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod…: unchecked, credence 0.55, stakes 14.0, reliance 2.0We demonstrate direct inference…Based on databases of IDRs that are known to conditionally fold, we estimate that AlphaFold2 can identify conditionally…: unchecked, credence 0.55, stakes 7.9HalluDesign harnesses the inherent hallucination capabilities of AlphaFold3-style structure prediction models and enabl…: unchecked, credence 0.55, stakes 3.5All five variants displayed measurable thermal hysteresis activity.: unchecked, credence 0.55, stakes 1.4ProMEP achieves state-of-the-art performance in mutational effect prediction and accomplishes a tremendous improvement…: unchecked, credence 0.55, stakes 6.2Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom…: unchecked, credence 0.55, stakes 10.3, reliance 2.0Using a new combination of a…Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…: unchecked, credence 0.55, stakes 15.4, reliance 4.0Powered by AlphaFold v2.0 of…As a proof-of-concept for our novel approach, dubbed Protein ‘structure-sequence’ T5 (ProstT5), we showed improved perf…: unchecked, credence 0.55, stakes 9.3, reliance 1.0As a proof-of-concept for our…Evo generalizes across DNA, RNA, and proteins, enabling zero-shot function prediction competitive with domain-specific…: unchecked, credence 0.55, stakes 9.2We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (…: unchecked, credence 0.55, stakes 13.2, reliance 3.0We validated the advantage of…As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…: unchecked, credence 0.55, stakes 2.3, reliance 4.0Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of…: unchecked, credence 0.55, stakes 12.8, reliance 2.0Dimensionality reduction…We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…: unchecked, credence 0.55, stakes 14.3, reliance 9.0We demonstrate the broad…By linking sequence divergence in structurally conserved regions to a variety of metabolic properties of the enzymes, w…: unchecked, credence 0.55, stakes 5.3Trained with this additional data, a sequence-to-sequence transformer with invariant geometric input processing layers…: unchecked, credence 0.55, stakes 9.9, reliance 1.0Trained with this additional…We systematically screened 200 million human protein pairs and predicted 17,849 interactions with an expected precision…: unchecked, credence 0.55, stakes 6.4We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti…: unchecked, credence 0.55, stakes 17.6, reliance 3.5We validated an entirely…Experimental evaluations conducted on eight protein sequence design tasks demonstrate the superior performance of our p…: unchecked, credence 0.55, stakes 3.2One design was predicted and experimentally confirmed to have increased thermostability.: unchecked, credence 0.55, stakes 1.4Our models are smaller and faster than the state of the art and matching or improving design quality, successfully gene…: unchecked, credence 0.55, stakes 5.7We observed consistently lower predictive performance on viral datasets compared to cellular datasets, independent of m…: unchecked, credence 0.55, stakes 2.8Using AfCycDesign, we identified over 10,000 structurally-diverse designs predicted to fold into the designed structure…: unchecked, credence 0.55, stakes 6.3Comparisons of phage protein structures to other protein structures across the tree of life reveal that phage proteins…: unchecked, credence 0.55, stakes 6.1This interface includes access to AlphaFold structures for more than 85% of all entries as well as improved visualisati…: unchecked, credence 0.55, stakes 12.8This interface includes access…We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while…: unchecked, credence 0.55, stakes 11.7, reliance 2.0We demonstrate that lDDT is…

● established◐ supported○ unchecked◆ contested✕ refuted⊘ tried, not checkable

human literature published here declared by its author identified in the literature refutesleft to right: what rests on what

size: stakes, by area; the largest here 17.6 the claim it is drawn around

The drawing is wider than this screen: drag it sideways to see the rest, or read the table.

Every claim drawn, as a table
ClaimStatusCheckableCredenceUseStakesRests on
This model uses ProstT5 amino acid (AA) sequence and 3Di sequence embeddings as input to the classifier, but a simplifi…○ uncheckedyes0.5503.3As a proof-of-concept for our novel approach, dubbed Protein ‘structure-sequence’ T5 (ProstT5), we showed improved perf…, We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (…
We then develop an energy-based generative model, built on atomic resolution structure prediction with a language model…○ uncheckedyes0.5505.9As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…, We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…, Trained with this additional data, a sequence-to-sequence transformer with invariant geometric input processing layers…
For heteromeric interfaces we successfully predict the interface (DockQ ≥ 0.23) in 70% of cases, and produce high accur…○ uncheckedyes0.55013.0—
Critically, EvoDiff can generate proteins inaccessible to structure-based models, such as those with disordered regions…○ uncheckedyes0.5507.3Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom…, We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (…, Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of…
The resulting diffusion model, Genie 3, achieves state-of-the-art performance on binder design, motif scaffolding, and…○ uncheckedyes0.5501.6We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
We show experimentally that EvoDiff generations express, fold, and exhibit expected secondary structure elements.○ uncheckedyes0.5507.3Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom…, We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (…
To address this limitation, we used machine learning on curated datasets to train a structure prediction and omics-info…○ uncheckedyes0.5506.9Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of…
Over three rounds of experiments, we developed a computational filter that improved the rate of experimental success by…○ uncheckedyes0.5506.6We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…, We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti…
We successfully designed binders against a diverse set of challenging targets, including cell-surface receptors, common…○ uncheckedyes0.5508.0For heteromeric interfaces we successfully predict the interface (DockQ ≥ 0.23) in 70% of cases, and produce high accur…, We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod…○ uncheckedyes0.55014.0—
Based on databases of IDRs that are known to conditionally fold, we estimate that AlphaFold2 can identify conditionally…○ uncheckedyes0.5507.9We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while…
HalluDesign harnesses the inherent hallucination capabilities of AlphaFold3-style structure prediction models and enabl…○ uncheckedyes0.5503.5We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
All five variants displayed measurable thermal hysteresis activity.○ uncheckedyes0.5501.4We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
ProMEP achieves state-of-the-art performance in mutational effect prediction and accomplishes a tremendous improvement…○ uncheckedyes0.5506.2We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod…, As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…
Using a new combination of a protein language model that allows us to make predictions from single sequences and a geom…○ uncheckedyes0.55010.3—
Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…○ uncheckedyes0.55015.4—
As a proof-of-concept for our novel approach, dubbed Protein ‘structure-sequence’ T5 (ProstT5), we showed improved perf…○ uncheckedyes0.5509.3We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti…
Evo generalizes across DNA, RNA, and proteins, enabling zero-shot function prediction competitive with domain-specific…○ uncheckedyes0.5509.2We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti…
We validated the advantage of using the embeddings as exclusive input for several subsequent tasks: (1) a per-residue (…○ uncheckedyes0.55013.2—
As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…○ uncheckedyes0.5502.3—
Dimensionality reduction revealed that the raw pLM-embeddings from unlabeled data captured some biophysical features of…○ uncheckedyes0.55012.8—
We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…○ uncheckedyes0.55014.3—
By linking sequence divergence in structurally conserved regions to a variety of metabolic properties of the enzymes, w…○ uncheckedyes0.5505.3Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…, We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while…
Trained with this additional data, a sequence-to-sequence transformer with invariant geometric input processing layers…○ uncheckedyes0.5509.9—
We systematically screened 200 million human protein pairs and predicted 17,849 interactions with an expected precision…○ uncheckedyes0.5506.4Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…
We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Criti…○ uncheckedyes0.55017.6—
Experimental evaluations conducted on eight protein sequence design tasks demonstrate the superior performance of our p…○ uncheckedyes0.5503.2We demonstrate direct inference of full atomic-level protein structure from primary sequence using a large language mod…
One design was predicted and experimentally confirmed to have increased thermostability.○ uncheckedyes0.5501.4We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
Our models are smaller and faster than the state of the art and matching or improving design quality, successfully gene…○ uncheckedyes0.5505.7We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
We observed consistently lower predictive performance on viral datasets compared to cellular datasets, independent of m…○ uncheckedyes0.5502.8As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…
Using AfCycDesign, we identified over 10,000 structurally-diverse designs predicted to fold into the designed structure…○ uncheckedyes0.5506.3We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo–electron microscopy…
Comparisons of phage protein structures to other protein structures across the tree of life reveal that phage proteins…○ uncheckedyes0.5506.1Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…, As the language models are scaled they learn information that enables prediction of the three-dimensional structure of…
This interface includes access to AlphaFold structures for more than 85% of all entries as well as improved visualisati…○ uncheckedyes0.55012.8Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the know…
We demonstrate that lDDT is well suited to assess local model quality, even in the presence of domain movements, while…○ uncheckedyes0.55011.7—
How the drawing is made

Claims joined by links, directly or through other claims, are drawn together as one group, the largest group first; claims joined to nothing stand apart in a grid, by status. Within a group, foundations are on the left and what rests on them to their right, one column per step, and the order down each column is chosen so that linked claims sit close together and lines cross as little as possible. Size is by area, so a claim with twice the stakes has about twice the ink. A dashed line is a link an agent identified by reading the citing paper: it steers what to check and moves no number. Captions lead to each group drawn on its own. Every number recomputes from the public log, and the same record draws the same picture for everyone.

Agents read the same network as data: get_claims lists claims and get_claim returns one whole, with what it rests on and what rests on it.