Abstract Human fragile sites are genomic regions that are susceptible to DNA damage and breakage, especially under replication stress. They are strongly associated with genomic instability and play a critical role in diseases such as cancer and neurological disorders. Accurate detection of fragile sites can therefore advance our understanding of disease mechanisms and inform therapeutic strategies. However, only a few computational methods exist for predicting DNA fragility, and they typically rely on extensive feature engineering prior to inference. Genomic language models (gLMs), trained directly on raw DNA sequences, have recently emerged as powerful tools for diverse biological prediction tasks. Here, we explore the application of gLMs to predict fragile sites in the human reference genome. We first constructed a balanced training dataset consisting of 3073 gene sequences located within manually curated fragile sites from the HumanCFS database and an equal number of negative sequences sampled from random genomic regions. Using a 70–10-20 train-validation-test split of this dataset, we evaluated the performance of three gLMs (Caduceus, HyenaDNA, and Nucleotide Transformer), together with a convolutional neural network, to predict fragile sites. Our results show that fragile sites exhibit distinctive nucleotide-level patterns that can be effectively learned by deep neural networks, particularly those modelling long-range interactions. Beyond predictive accuracy and efficiency, interpretability analyses and embedding visualisations indicate that these models capture biologically meaningful patterns. Future work will expand this framework with larger datasets and extend its scope to predicting rearrangement hotspots and genome-wide DNA fragility.

SUMMARY: Simulating realistic clonal dynamics of tumors is an important topic in cancer genomics. Here, we present Phylogeny guided Simulator for Tumor Evolution, a tool that can simulate different types of tumor samples including single sector, multi-sector bulk tumor as well as single-cell tumor data under a wide range of evolutionary trajectories. Phylogeny guided Simulator for Tumor Evolution provides an efficient tool for understanding clonal evolution of cancer. AVAILABILITY AND IMPLEMENTATION: PSiTE is implemented in Python and is available at https://github.com/hchyang/PSiTE. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Genetic material is transferred in a non-reproductive manner across species more frequently than commonly thought, particularly in the bacteria kingdom. On one hand, extant genomes are thus more properly considered as a fusion product of both reproductive and non-reproductive genetic transfers. This has motivated researchers to adopt phylogenetic networks to study genome evolution. On the other hand, a gene's evolution is usually tree-like and has been studied for over half a century. Accordingly, the relationships between phylogenetic trees and networks are the basis for the reconstruction and verification of phylogenetic networks. One important problem in verifying a network model is determining whether or not certain existing phylogenetic trees are displayed in a phylogenetic network. This problem is formally called the tree containment problem. It is NP-complete even for binary phylogenetic networks. RESULTS: We design an exponential time but efficient method for determining whether or not a phylogenetic tree is displayed in an arbitrary phylogenetic network. It is developed on the basis of the so-called reticulation-visible property of phylogenetic networks. AVAILABILITY AND IMPLEMENTATION: A C-program is available for download on http://www.math.nus.edu.sg/∼matzlx/tcp_package CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.