Biology subjects

Akiva, E.

Publications and source records attributed to Akiva, E..

5 recordsLinked to original sources

bioRxiv2023

PiNUI: A Dataset of Protein-Protein Interactions for Machine Learning

We introduce a novel dataset named PiNUI: Protein Interactions with Nearly Uniform Imbalance. PiNUI is a dataset of Protein-Protein Interactions (PPI) specifically designed for Machine Learning (ML) applications that offer a higher degree of representativeness of real-world PPI tasks compared to existing ML-ready PPI datasets. We achieve such by increasing the data size and quality, and minimizing the sampling bias of negative interactions. We demonstrate that models trained on PiNUI almost always outperform those trained on conventional PPI datasets when evaluated on various general PPI tasks using external test sets. PiNUI is available here.

bioinformatics↗

bioRxiv2022

Improving Protein Subcellular Localization Prediction with Structural Prediction & Graph Neural Networks

The majority of biological functions are carried out by proteins. Proteins perform their roles only upon arrival to their target location in the cell, hence elucidating protein subcellular localization is essential for better understanding their function. The exponential growth in genomic information and the high cost of experimental validation of protein localization call for the development of predictive methods. We present a method that improves subcellular localization prediction for proteins based on their sequence by leveraging structure prediction and Graph Neural Networks. We demonstrate how Language Models, trained on protein sequences, and Graph Neural Networks, trained on proteins 3D structures, are both efficient approaches for this task. They both learn meaningful, yet different representations of proteins; hence, ensembling them outperforms the reigning state of the art method. Our architecture improves the localization prediction performance while being lighter and more cost-effective.

bioinformatics↗

bioRxiv2021

Protein Organization with Manifold Exploration and Spectral Clustering

We present a method to provide a biologically meaningful representation of the space of protein sequences. While billions of protein sequences are available, organizing this vast amount of information into functional categories is daunting, time-consuming and incomplete. We present our unsupervised approach that combines Transformer protein language models, UMAP graphs, and spectral clustering to create meaningful clusters in the protein spaces. To demonstrate the meaningfulness of the clusters, we show that they preserve most of the signal present in a dataset of manually curated enzyme protein families.

bioinformatics↗

bioRxiv2020

Revealing enzyme functional architecture via high-throughput microfluidic enzyme kinetics

Systematic and extensive investigation of enzymes is needed to understand their extraordinary efficiency and meet current challenges in medicine and engineering. We present HT-MEK, a microfluidic platform for high-throughput expression, purification, and characterization of >1500 enzyme variants per experiment. For 1036 mutants of the alkaline phosphatase PafA, we performed >670,000 reactions to determine >5000 kinetic and physical constants for multiple substrates and inhibitors. These constants allowed us to uncover extensive kinetic partitioning to a misfolded state and isolate catalytic effects, revealing spatially contiguous "regions" of residues linked to particular aspects of function. These regions included active-site proximal residues but also extended to the enzyme surface, providing a map of underlying architecture that could not be derived from existing approaches. HT-MEK, using direct and coupled fluorescent assays, has future applications to a wide variety of problems ranging from understanding molecular mechanisms to medicine to engineering and design. One Sentence SummaryHT-MEK, a microfluidic platform for high-throughput, quantitative biochemistry, reveals enzyme architectures shaping function.

biochemistry↗

bioRxiv2020

The evolution of human SIRT1 for multi-specificity is enabled by active-site adaptation

Many enzymes that catalyze protein post-translational modifications (PTM) can specifically modify multiple target proteins. However, little is known regarding the molecular basis and evolution of multi-specificity in these enzymes. Here, we used bioinformatics and experimental approaches to investigate the molecular basis and evolution of multi-specificity in the sirtuin-1 (SIRT1) deacetylase. Guided by bioinformatics analysis of SIRT1 orthologues and substrates, we identified and examined key mutations that have occurred during the evolution of human SIRT1. We found that while these mutants maintain high activity toward conserved histone substrates they exhibit a dramatic loss of activity toward acetylated p53 protein that appeared only in Metazoa. These results demonstrate that active site substitutions in SIRT1 are essential to enable the inclusion of p53 substrate. Our results are consistent with a model in which promiscuous ancestral activity can provide an evolutionary starting point from which multispecificity toward a number of cellular substrates can develop.

biochemistry↗