Search bioRxiv⌕ Search

Biology subjects

Taghon, G.

Publications and source records attributed to Taghon, G..

2 recordsLinked to original sources

Towards interoperable modeling of toehold-mediated strand exchange circuits across DNA nanotechnology and engineering biology

Originally developed for DNA nanotechnology, toehold-mediated strand exchange (TMSE) circuits are gaining traction in synthetic biology due to their high programmability, seamless integration with biological components, and robust operation across diverse environments and cell types. However, while forward-engineering in synthetic biology has benefited from automated genetic circuit modeling pipelines, there is currently a lack of accessible, automated tools for the mechanistic modeling of TMSE circuits integrated with these systems, hindering the development of new biotechnologies. The TMSE-BioCRNpyler Library allows TMSE molecules to be transcribed RNAs or fixed-concentration nucleic acids while leveraging existing BioCRNpyler features, such as upstream transcription regulation and downstream gene regulation. We demonstrate this librarys applicability by modeling published applications of TMSE circuits spanning a wide range of applications, including simple in vitro reactions, cell-free biosensors, and in vivo microbial and mammalian systems. Additionally, we validated that models compiled using the TMSE-BioCRNpyler Library produced results with < 0.2 % relative error compared to multiple models of TMSE previously developed in the literature. Finally, to streamline interoperability with existing models, we developed txt2biocrnpyler. This accompanying tool converts chemical reaction networks from the literature into a BioCRNpyler-ready source script and a Systems Biology Markup Language XML file -- a standard data format for sharing and simulating biological models. The TMSE-BioCRNpyler Library serves as a powerful new resource for the rational, automated design of molecular information processing systems.

bioengineering↗

SimpleFold-Turbo: Adaptive Inference Caching Yields 14-fold Acceleration of Flow-Matching Protein Structure Prediction

We apply TeaCache, an adaptive caching technique from video diffusion to SimpleFolds flow-matching protein structure prediction and achieve (9 to 14)-fold inference speedups with negligible quality loss. We determine that flow matchings near-linear generative trajectories make consecutive neural-network evaluations highly redundant. At a low redundancy threshold, SimpleFold-Turbo (SF-T) skips {approx} 93 % of forward passes while preserving near-baseline template modeling (TM)-scores across 300 structurally diverse CATH domains and all six SimpleFold model sizes (100 million to 3 billion parameters), at compute budgets where log-uniform step-skipping collapses. Speedup scales with model size because caching overhead is constant while per-step cost grows, and a general three-phase skip pattern emerges independent of protein size or fold. SF-T requires no retraining, no weight modification, and no MSA server dependencies. We release SF-T as fully open-source software enabling thousands of structure predictions per hour on commodity hardware.

bioinformatics↗