Post
2230
Latest OpenAI Parameter Golf Competition Training Run BPB (<1K steps on a single 4090) See: ToricBLM, ToricGT, and TropicalGT methods
- 1 reply
Amelie Schreiber
AmelieSchreiber
AI & ML interests
Diffusion and flow matching models for proteins, small molecules, DNA, and RNA, protein language models, machine translation, equivariant attention mechanisms, LoRA, and QLoRA.
Recent Activity
updated a dataset 17 days ago
AmelieSchreiber/toricblm-dataset-state-toricblm-structure-priority-balanced-3day-20260709t185034z-epoch-002 published a dataset 17 days ago
AmelieSchreiber/toricblm-dataset-state-toricblm-structure-priority-balanced-3day-20260709t185034z-epoch-002 updated a dataset 17 days ago
AmelieSchreiber/toricblm-dataset-state-toricblm-structure-priority-balanced-3day-20260709t185034z-epoch-001Organizations
posted an update about 1 month ago
posted an update about 1 month ago
Post
170
In algebraic geometry, embedding tropical varieties into toric varieties relies on the process of tropicalization and the use of toroidal embeddings. Rather than embedding as geometric spaces, tropical varieties are embedded as polyhedral fans or cone complexes within the fans of target toric varieties to study compactifications and intersection theory.
Toric Varieties: Geometric varieties constructed from combinatorial data known as fans (collections of polyhedral cones). Tropicalization: A procedure that maps an algebraic variety X (typically a subvariety of an algebraic torus T) to a polyhedral complex in a real vector space via valuations. Toric Closures: When you embed a tropical variety in a compact toric variety, the boundary points correspond to limits of points in the variety going towards the boundary of the torus. Embedding Methodologies...Toric Compactification: Any tropical variety (a rational polyhedral fan) can be compactified by embedding it into a complete toric variety. The fan of the toric variety is chosen to contain the tropical variety as a subcomplex, which encodes the combinatorics of the compactification. Toroidal Embeddings: For varieties that are not strict toric varieties, they can often be mapped into toric varieties using the theory of toroidal embeddings. This involves equipping the tropicalized spaces (cone complexes) with balancing conditions and intersection theories.
Toric Varieties: Geometric varieties constructed from combinatorial data known as fans (collections of polyhedral cones). Tropicalization: A procedure that maps an algebraic variety X (typically a subvariety of an algebraic torus T) to a polyhedral complex in a real vector space via valuations. Toric Closures: When you embed a tropical variety in a compact toric variety, the boundary points correspond to limits of points in the variety going towards the boundary of the torus. Embedding Methodologies...Toric Compactification: Any tropical variety (a rational polyhedral fan) can be compactified by embedding it into a complete toric variety. The fan of the toric variety is chosen to contain the tropical variety as a subcomplex, which encodes the combinatorics of the compactification. Toroidal Embeddings: For varieties that are not strict toric varieties, they can often be mapped into toric varieties using the theory of toroidal embeddings. This involves equipping the tropicalized spaces (cone complexes) with balancing conditions and intersection theories.
posted an update 2 months ago
Post
248
# New Spin on Old Methods
---
- https://huggingface.co/blog/AmelieSchreiber/toricgt
- https://huggingface.co/collections/AmelieSchreiber/toricgt
---
~A
---
- https://huggingface.co/blog/AmelieSchreiber/toricgt
- https://huggingface.co/collections/AmelieSchreiber/toricgt
---
~A
replied to their post almost 2 years ago
<3
replied to their post almost 2 years ago
Aww thank you!
posted an update about 2 years ago
Post
6083
A list of my favorite articles I've written on Hugging Face (the others are kind of mid imo):
https://huggingface.co/blog/AmelieSchreiber/a-new-era-of-enzyme-engineering
https://huggingface.co/blog/AmelieSchreiber/protein-optimization-and-design
https://huggingface.co/blog/AmelieSchreiber/rfdiffusion-potentials
https://huggingface.co/blog/AmelieSchreiber/mutation-scoring
https://huggingface.co/blog/AmelieSchreiber/esmbind
https://huggingface.co/blog/AmelieSchreiber/esm2-ptm
https://huggingface.co/blog/AmelieSchreiber/directed-evolution-with-esm2
https://huggingface.co/blog/AmelieSchreiber/intrinsic-dimension-of-proteins
https://huggingface.co/blog/AmelieSchreiber/plm-persistent-homology-msa-replacement
https://huggingface.co/blog/AmelieSchreiber/faster-pha
https://huggingface.co/blog/AmelieSchreiber/a-new-era-of-enzyme-engineering
https://huggingface.co/blog/AmelieSchreiber/protein-optimization-and-design
https://huggingface.co/blog/AmelieSchreiber/rfdiffusion-potentials
https://huggingface.co/blog/AmelieSchreiber/mutation-scoring
https://huggingface.co/blog/AmelieSchreiber/esmbind
https://huggingface.co/blog/AmelieSchreiber/esm2-ptm
https://huggingface.co/blog/AmelieSchreiber/directed-evolution-with-esm2
https://huggingface.co/blog/AmelieSchreiber/intrinsic-dimension-of-proteins
https://huggingface.co/blog/AmelieSchreiber/plm-persistent-homology-msa-replacement
https://huggingface.co/blog/AmelieSchreiber/faster-pha
- 4 replies
replied to their post over 2 years ago
It was unexpected. I believe finetuning may be in order and EvoDiff is not really intended to generate binders out-of-the-box anyway. Still though, the changes in tertiary structure for the generated protein are puzzling and interesting. How does an alpha-helix with an some higher pLDDT scores (middle picture of the unbound generated binder) turn into beta-sheets with low pLDDT scores in the presence of the target protein (picture on the right)? I find that strange and interesting. Is this a more extreme example of sampling multiple conformations of proteins that are fold-switching? (see https://www.biorxiv.org/content/10.1101/2023.12.16.571997v1)
posted an update over 2 years ago
Post
Currently attempting to hack EvoDiff to generate binders for target proteins with some interesting results. The generated binders tend to change conformation, sometimes drastically, when bound to the target proteins compared to their unbound states. Below is the target protein with an IDR linker, the generated binder, and the binder bound to the target protein with the IDR linker structure as predicted by ESMFold. Notice how the binder goes from being a solid alpha-helix, to being beta-sheets (in orange). That's quite a change in tertiary structure!
- 3 replies