MOTIVATION: Gene Set Enrichment Analysis (GSEA) is widely used to interpret DNA methylation data by associating differentially methylated sites with biological pathways. However, existing GSEA methods struggle with several challenges in methylation data, including probe dependency, probe number bias, and the complexity of gene-probe mapping. These limitations can lead to biased enrichment results, reduced statistical power, and computational inefficiencies. RESULTS: We introduce gsGene and gsPG, two novel GSEA methods specifically designed for DNA methylation data. gsGene aggregates association signals at gene level while correcting for probe dependency and probe number bias, enabling more biologically meaningful enrichment analysis. gsPG takes a different approach by conducting gene set enrichment using summary statistics for independent probe groups based on gene annotation, mitigating biases from multi-mapping probes. Both methods improve computational efficiency, enhance statistical power, and effectively control type I error rates. Comprehensive evaluations in two large datasets demonstrate superior performance compared to existing methods. Furthermore, we propose a novel beta distribution fitting strategy to improve enrichment P-value estimation, providing a computationally efficient alternative to traditional permutation-based gene set methods. AVAILABILITY AND IMPLEMENTATION: These methods are implemented in the R package dmGsea, which is freely available on GitHub and Bioconductor (DOI: 10.18129/B9.bioc.dmGsea). The package supports Illumina 450K, EPIC, and mouse methylation arrays and can be extended to other omics data with user-provided probe-to-gene mapping annotations.

MOTIVATION: DNA methylation-based predictors of various biological metrics have been widely published and are becoming valuable tools in epidemiologic studies of epigenetics and personalized medicine. However, generating these predictors from original source software and web servers is complex and time consuming. Furthermore, different predictors were often derived based on data from different types of arrays, where array differences and batch effects can make predictors difficult to compare across studies. RESULTS: We integrate these published methods into a single R function to produce 158 previously published predictors for chronological age, biological age, exposures, lifestyle traits and serum protein levels using both classical and principal component-based methods. To mitigate batch and array differences, we also provide a modified RCP method (ref-RCP) that normalize input DNA methylation data to reference data prior to estimation. Evaluations in real datasets show that this approach improves estimate precision and comparability across studies. AVAILABILITY AND IMPLEMENTATION: The function was included in software package ENmix, and is freely available from Bioconductor website (https://www.bioconductor.org/packages/release/bioc/html/ENmix.html).

Multiple Object Tracking (MOT) has been a useful yet challenging task in many real-world applications such as video surveillance, intelligent retail, and smart city. The challenge is how to model long-term temporal dependencies in an efficient manner. Some recent works employ Recurrent Neural Networks (RNN) to obtain good performance, which, however, requires a large amount of training data. In this paper, we proposed a novel tracking method that integrates the auto-tuning Kalman method for prediction and the Gated Recurrent Unit (GRU), and achieves a near-optimum with a small amount of training data. Experimental results show that our new algorithm can achieve competitive performance on the challenging MOT benchmark, with higher efficiency and more robustness compared to the state-of-the-art RNN-based online MOT algorithms.

SUMMARY: ipDMR is an R software tool for identification of differentially methylated regions (DMRs) using auto-correlated P-values for individual CpGs from epigenome-wide association analysis using array or bisulfite sequencing data. It summarizes P-values for adjacent CpGs, identifies association peaks and then extends peaks to find boundaries of DMRs. ipDMR uses BED format files as input and is easy to use. Simulations guided by real data found that ipDMR outperformed current available methods and provided slightly higher true positive rates and much lower false discovery rates. AVAILABILITY AND IMPLEMENTATION: ipDMR is available at https://bioconductor.org/packages/release/bioc/html/ENmix.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Unmanned Aerial Vehicles (UAVs), better known as drones, have significantly advanced fields such as aerial surveillance, military reconnaissance, cadastral surveying, disaster monitoring, and delivery services. However, UAVs rely on civilian (unauthenticated) GPS for navigation which can be trivially spoofed.

This paper studies the problem of Person Re-Identification (ReID) for large-scale applications. Recent research efforts have been devoted to building complicated part models, which introduce considerably high computational cost and memory consumption, inhibiting its practicability in large-scale applications. This paper aims to develop a novel learning strategy to find efficient feature embeddings while maintaining the balance of accuracy and model complexity. More specifically, we find by enhancing the classical triplet loss together with cross-entropy loss, our method can explore the hard examples and build a discriminant feature embedding yet compact enough for large-scale applications. Our method is carried out progressively using Bayesian optimization, and we call it the Progressive Learning Algorithm (PLA). Extensive experiments on three large-scale datasets show that our PLA is comparable or better than the-state-of-the-arts. Especially, on the challenging Market-ISOI dataset, we achieve Rank-1=94.7%/mAP=89.4% while saving at least 30 % parameters than strong part models.

Multi-label hashing is a new research topic in image retrieval. As images are usually associated with multiple semantic labels, there is multi-level semantic similarity such as very similar, normally similar and dissimilar among multi-label images. In order to obtain the multi-level semantic similarity, this letter constructs a hypergraph in label space by creating a hyperedge for each semantic label and including all images annotated with a common label into one hyperedge. In this way, the number of common hyperedges shared by the vertices in hypergraph can be used to encode the high-order semantic relations among multiple images. Considering the useful similarity information hidden in the instance space, a kNN graph in instance space is further constructed. By learning from both the hypergraph and kNN graph with spectral learning strategy, a graph regularized deep discrete hashing is developed which updates graph regularized binary codes and deep neural network based robust features iteratively in a discrete optimization framework. The results in comparison with nine state-of-the-art hashing methods on two multi-label image datasets such as MIRFLICKR-25 K and NUS-WISE demonstrate its effectiveness.

MOTIVATION: The Illumina HumanMethylation450 BeadChip has been extensively utilized in epigenome-wide association studies. This array and its successor, the MethylationEPIC array, use two types of probes-Infinium I (type I) and Infinium II (type II)-in order to increase genome coverage but differences in probe chemistries result in different type I and II distributions of methylation values. Ignoring the difference in distributions between the two probe types may bias downstream analysis. RESULTS: Here, we developed a novel method, called Regression on Correlated Probes (RCP), which uses the existing correlation between pairs of nearby type I and II probes to adjust the beta values of all type II probes. We evaluate the effect of this adjustment on reducing probe design type bias, reducing technical variation in duplicate samples, improving accuracy of measurements against known standards, and retention of biological signal. We find that RCP is statistically significantly better than unadjusted data or adjustment with alternative methods including SWAN and BMIQ. AVAILABILITY: We incorporated the method into the R package ENmix, which is freely available from the Bioconductor website (https://www.bioconductor.org/packages/release/bioc/html/ENmix.html). CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: 5-Methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) are important epigenetic regulators of gene expression. 5mC and 5hmC levels can be computationally inferred at single base resolution using sequencing or array data from paired DNA samples that have undergone bisulfite and oxidative bisulfite conversion. Current estimation methods have been shown to produce irregular estimates of 5hmC level or are extremely computation intensive. RESULTS: We developed an efficient method oxBS-MLE based on binomial modeling of paired bisulfite and oxidative bisulfite data from sequencing or array analysis. Evaluation in several datasets showed that it outperformed alternative methods in estimate accuracy and computation speed. AVAILABILITY AND IMPLEMENTATION: oxBS-MLE is implemented in Bioconductor package ENmix. CONTACT: [email protected] information: Supplementary data are available at Bioinformatics online.