Search for: All records

This research presents an enhanced Graph Attention Convolutional Neural Network (GAT) tailored for the analysis of open-source package vulnerability remediation. By meticulously examining control flow graphs and implementing node centrality metrics—specifically, degree, norm, and closeness centrality—our methodology identifies and evaluates changes resulting from vulnerability fixes in nodes, thereby predicting the ramifications of dependency upgrades on application workflows. Empirical testing on diverse datasets reveals that our model challenges established paradigms in software security, showcasing its efficacy in delivering comprehensive insights into code vulnerabilities and contributing to advancements in cybersecurity practices. This study delineates a strategic framework for the development of sustainable monitoring systems and the effective remediation of vulnerabilities in open-source software. 

ABSTRACT The evolution of oxygenic photosynthesis in the Cyanobacteria was one of the most transformative events in Earth history, eventually leading to the oxygenation of Earth’s atmosphere. However, it is difficult to understand how the earliest Cyanobacteria functioned or evolved on early Earth in part because we do not understand their ecology, including the environments in which they lived. Here, we use a cutting-edge bioinformatics tool to survey nearly 500,000 metagenomes for relatives of the taxa that likely bookended the evolution of oxygenic photosynthesis to identify the modern environments in which these organisms live. Ancestral state reconstruction suggests that the common ancestors of these organisms lived in terrestrial (soil and/or freshwater) environments. This restricted distribution may have increased the lag between the evolution of oxygenic photosynthesis and the oxygenation of Earth’s atmosphere.IMPORTANCECyanobacteria generate oxygen as part of their metabolism and are responsible for the rise of oxygen in Earth’s atmosphere over two billion years ago. However, we do not know how long this process may have taken. To help constrain how long this process would have taken, it is necessary to understand where the earliest Cyanobacteria may have lived. Here, we use a cutting-edge bioinformatics tool called branch water to examine the environments where modern Cyanobacteria and their relatives live to constrain those inhabited by the earliest Cyanobacteria. We find that these species likely lived in non-marine environments. This indicates that the rise of oxygen may have taken longer than previously believed. 

The first compilations of Proterozoic eukaryote diversity, published in the 1980s showed a dramatic peak in the Tonian Period (1000–720 Ma), interpreted as the initial radiation of eukaryotes in the marine realm. Over the decades, new discoveries filled in the older part of the record and the peak diminished, but the idea of a Tonian radiation of eukaryotes has remained strong, and is now widely accepted as fact. We present a new diversity compilation based on 181 species and 713 species occurrences from 145 formations ranging in age from 1890 Ma to 720 Ma and find a significant increase in diversity in the Tonian. However, we also find that the number of eukaryotic species through time is highly correlated with the number of formations in our dataset (i.e. eukaryote-bearing formations) through time. This correlation is robust to interpretations of eukaryote affinity, bin size, and bin boundaries. We also find that within-assemblage diversity—a measure thought to circumvent sampling bias—is related to the number of eukaryote-bearing formations through time. Biomarkers show a similar pattern to body fossils, where the rise of eukaryotic biosignatures correlates with increased sampling. We find no evidence that the proportion of eukaryote-bearing versus all fossiliferous formations changed through the Proterozoic, as might be expected if the correlation reflected an increase in eukaryote diversity driving an increase in the number of eukaryote-bearing formations. Although the correlation could reflect a common cause such as changes in sea level driving both diversification and an increase in sedimentary rock volume, we favor the explanation that the pattern of early eukaryote diversity is driven by variations in paleontological sampling. 

Bose-Einstein condensation of excitons, with its potential for frictionless energy transport, has recently been observed in materials at low temperatures. Here, we show that partial exciton condensation plays a significant role in the 18-chromophore B850 ring of the light-harvesting complex 2 (LH2) in purple bacteria. Even in the single-excitation regime, we observe that excitonic entanglement across multiple sites exhibits signatures of exciton condensation in the particle-hole reduced density matrix—a partial exciton condensate. Crucially, we find that, by distributing the exciton across multiple sites of the ring, the exciton-condensate-like state sets favorable conditions for enhanced energy transfer, both before and after decoherence. Surprisingly, this discovery reveals that excitonic condensation, previously thought to require extreme conditions, can occur in a partial form in biological systems under ambient conditions, providing new insight into energy transport. These results additionally bring new insight into the long-standing debate on quantum versus classical mechanisms in photosynthetic light harvesting by showing that quantum coherence, in the form of a partial exciton condensate, indirectly initializes subsequent classical transfer. Our findings not only deepen our understanding of quantum coherence in light harvesting but also suggest design principles for materials capable of leveraging excitonic entanglement for efficient energy transport. Published by the American Physical Society2025 

This paper demonstrates the utility of organized numerical representations of genes in research involving flat string gene formats (i.e., FASTA/FASTQ5). By assigning a unique vector embedding to each short sequence, it is possible to more efficiently cluster and improve upon compression performance for the string representations of cDNA libraries. Furthermore, by studying alternative coordinate vector embeddings trained on the context of codon triplets, we can demonstrate clustering based on amino acid properties. Employing this sequence embedding method to encode barcodes and cDNA sequences, we can improve the time complexity of similarity searches. By pairing vector embeddings with an algorithm that determines the vector proximity in Euclidean space, this approach enables quicker and more flexible sequence searches.