Matko Glunčić

Topic modeling is a fundamental technique in natural language processing used to uncover latent themes in large text corpora, yet existing approaches struggle to jointly achieve interpretability, semantic coherence, and scalability. Classical probabilistic models such as LDA and NMF rely on bag-of-words assumptions that obscure contextual meaning, while embedding-based methods (e.g., BERTopic, Top2Vec) improve coherence at the expense of diversity and stability. Prompt-based frameworks (e.g., TopicGPT) enhance interpretability but remain sensitive to prompt design and are computationally costly on large datasets. This study introduces VISTA (Vector-Similarity Topic Analysis), a multi-view, hierarchical, and interpretable framework that integrates complementary document embeddings, mutual-nearest-neighbor hierarchical clustering with selective dimension analysis, and large language model (LLM)-based topic labeling enforcing hierarchical coherence. Experiments on three heterogeneous corpora—BBC News, BillSum, and a mixed U.S. Government agency news + Twitter dataset—show that VISTA consistently ranks among the top-performing models, achieving the highest C_UCI coherence and a strong balance between topic diversity and semantic consistency. Qualitative analyses confirm that VISTA identifies domain-relevant themes overlooked by probabilistic or prompt-based models. Overall, VISTA provides a scalable, semantically robust, and interpretable framework for topic discovery, bridging probabilistic, embedding-based, and LLM-driven paradigms in a unified and reproducible design.

Recent analyses have revealed that orangutan alpha satellite higher-order repeat (HOR) arrays in complete centromeres are composed of three to four distinct HOR blocks, each sharing only 80–90% sequence identity, thus forming a patchwork-quilt pattern of independent HOR expansions. In contrast, using our novel HOR-detection algorithm GRhor, we analyzed the complete Y chromosome centromere in orangutan and identified a highly ordered and complex alpha satellite 58mer superHOR array, comprising 67 HOR copies, including 46 highly identical canonical copies with a remarkably low divergence of only 0.25%. Given that the largest known human alpha satellite HOR is the 34mer on the Y chromosome, this novel 58mer structure qualifies as a superHOR. The canonical 58mer HOR contains only 44 distinct monomer types, with 14 types repeated within the unit, resulting in a unique five-row cascading organization. Such complexity is not detectable using standard HOR-searching tools employed in previous studies. Additionally, we identified a second, less pronounced 45mer cascading superHOR array with 0.81% divergence. For comparative purposes, we also detected a cascading 18mer HOR in gorilla and a Willard-type 28mer HOR in chimpanzee Y centromeres. Notably, preliminary genome-wide analysis in orangutan reveals other superHORs, including 84mer and 53mer arrays in chromosome 5; a 54mer in chromosome 10; a 51mer in chromosome 14; a 53mer in chromosome 15; and a 45mer in chromosome 22. These findings underscore the power of GRMhor in revealing highly structured and species-specific HOR architectures, with potential implications for centromere evolution and primate comparative genomics.

Background: Centromeric alpha satellite DNA is organized into higher-order repeats (HORs), whose precise structure is often difficult to resolve in standard genome assemblies. The recent telomere-to-telomere (T2T) assembly of the human genome enables complete analysis of centromeric regions, including the full structure of HOR arrays. Methods: We applied the novel high-precision GRMhor algorithm to the complete T2T-CHM13 assembly of human chromosome 21. GRMhor integrates global repeat map (GRM) and monomer distance (MD) diagrams to accurately identify, classify, and visualize HORs and their subfragments. Results: The analysis revealed a novel Cascading 11mer HOR array, in which each canonical HOR copy comprises 11 monomers belonging to 10 different monomer types. Subfragments with periodicities of 4, 7, 9, and 20 were identified within the array. A second, complex 23/25mer HOR array of mixed Willard’s/Cascading type was also detected. In contrast to the hg38 assembly, where a dominant 8mer and 33mer HOR were previously annotated, these structures were absent in the T2T-CHM13 assembly, highlighting the limitations of hg38. Notably, we discovered a novel 52mer HOR—the longest alpha satellite HOR unit reported in the human genome to date. Several subfragment repeats correspond to alphoid subfamilies previously identified using restriction enzyme digestion, but are here resolved with higher structural precision. Conclusions: Our findings demonstrate the power of GRMhor in resolving complex and previously undetected alpha satellite architectures, including the longest canonical HOR unit identified in the human genome. The precise delineation of superHORs, Cascading structures, and HOR subfragments provides unprecedented insight into the fine-scale organization of the centromeric region of chromosome 21. These results highlight both the inadequacy of earlier assemblies, such as hg38, and the critical importance of complete telomere-to-telomere assemblies for accurately characterizing centromeric DNA.

Background/Objectives: The ~1.6 kb NBPF repeat units in neuroblastoma breakpoint family (NBPF) genes are specific to humans and are associated with cognitive capacity in higher primates. While the number of NBPF monomers/Olduvai sequences in humans is approximately 2–3 times greater than in great apes, the difference in copy number values of canonical NBPF 3mer Higher-order repeats (HORs)/Olduvai triplets between humans and great apes is substantially larger. This study aims to analyze the organization and evolutionary significance of NBPF 3mer HORs/Olduvai triplets in fully sequenced primate genomes. Methods: We applied the global repeat map (GRM) algorithm to identify canonical and variant NBPF 3mer HORs/Olduvai triplets in the complete genomes of humans, chimpanzees, gorillas, and orangutans. The resulting monomer arrays were analyzed using the GRMhor algorithm to generate detailed schematic representations of NBPF HOR organization. Results: The analysis reveals a distinct difference in NBPF-related patterns among these primates, particularly in the number of tandemly organized canonical 3mer HORs/Olduvai triplets: 61 tandemly organized canonical NBPF 3mer HORs/Olduvai triplets in humans, compared to 0 in chimpanzees and orangutans, and 9 in gorillas. When considering only tandemly organized 3mer HORs/Olduvai triplets with more than three copies, the numbers adjust to 36 in humans and 0 in great apes. Furthermore, the divergence between individual NBPF monomers in humans and great apes is twice as high as that observed within great apes. Conclusions: These findings support the hypothesis that the tandem organization of NBPF 3mer HORs/Olduvai triplets plays a crucial role in enhancing cognitive capacity in humans compared to great apes, potentially providing a significant evolutionary advantage. This effect complements the impact of the increased number of individual NBPF monomers/Olduvai sequences, together contributing to a synergistic amplification effect.

From the recent genome assembly NHGRI_mPonAbe1-v2.0_NCBI (GCF_028885655.2) of orangutan chromosome 13, we computed the precise alpha satellite higher-order repeat (HOR) structure using the novel high-precision GRM2023 algorithm with Global Repeat Map (GRM) and Monomer Distance (MD) diagrams. This study rigorously identified alpha satellite HORs in the centromere of orangutan chromosome 13, discovering a novel 59mer HOR—the longest HOR unit identified in any primate to date. Additionally, it revealed the first intertwined sequence of three HORs, 18mer/27mer/45mer HORs, with a common aligned “backbone” across all HOR copies. The major 7mer HOR exhibits a Willard’s-type canonical copy, although some segments of the array display significant irregularities. In contrast, the 14mer HOR forms a regular Willard’s-type HOR array. Surprisingly, the GRM2023 high-precision analysis of chromosome 13 of human genome assembly T2T-CHM13v2.0 reveals the presence of only a 7mer HOR, despite both the orangutan and human genome assemblies being derived from whole genome shotgun sequences.

To precisely identify and analyze alpha-satellite higher-order repeats (HORs) in T2T-CHM13 assembly of human chromosome 3. From the recently sequenced complete T2T-CHM13 assembly of human chromosome 3, the precise alpha satellite HOR structure was ...

Unraveling the intricate centromere structure of human chromosomes holds profound implications, illuminating fundamental genetic mechanisms and potentially advancing our comprehension of genetic disorders and therapeutic interventions. This study rigorously identified and structurally analyzed alpha satellite higher-order repeats (HORs) within the centromere of human chromosome 15 in the complete T2T-CHM13 assembly using the high-precision GRM2023 algorithm. The most extensive alpha satellite HOR array in chromosome 15 reveals a novel cascading HOR, housing 429 15mer HOR copies, containing 4-, 7- and 11-monomer subfragments. Within each row of cascading HORs, all alpha satellite monomers are of distinct types, as in regular Willard’s HORs. However, different HOR copies within the same cascading 15mer HOR contain more than one monomer of the same type. Each canonical 15mer HOR copy comprises 15 monomers belonging to only 9 different monomer types. Notably, 65% of the 429 15mer cascading HOR copies exhibit canonical structures, while 35% display variant configurations. Identified as the second most extensive alpha satellite HOR, another novel cascading HOR within human chromosome 15 encompasses 164 20mer HOR copies, each featuring two subfragments. Moreover, a distinct pattern emerges as interspersed 25mer/26mer structures differing from regular Willard’s HORs and giving rise to a 34-monomer subfragment. Only a minor 18mer HOR array of 12 HOR copies is of the regular Willard’s type. These revelations highlight the complexity within the chromosome 15 centromeric region, accentuating deviations from anticipated highly regular patterns and hinting at profound information encoding and functional potential within the human centromere.

Motivation: Tandem monomeric units, integral components of eukaryotic genomes, form higher-order repeat (HOR) structures that play crucial roles in maintaining chromosome integrity and regulating gene expression and protein abundance. Given their significant influence on processes such as evolution, chromosome segregation, and disease, developing a sensitive and automated tool for identifying HORs across diverse genomic sequences is essential. Results: In this study, we applied the GRMhor (Global Repeat Map hor) algorithm to analyse the centromeric region of chromosome 20 in three individual human genomes, as well as in the centromeric regions of three higher primates. In all three human genomes, we identified six distinct HOR arrays, which revealed significantly greater differences in the number of canonical and variant copies, as well as in their overall structure, than would be expected given the 99.9% genetic similarity among humans. Furthermore, our analysis of higher primate genomes, which revealed entirely different HOR sequences, indicates a much larger genomic divergence between humans and higher primates than previously recognized. These results underscore the suitability of the GRMhor algorithm for studying specificities in individual genomes, particularly those involving repetitive monomers in centromere structure, which is essential for proper chromosome segregation during cell division, while also highlighting its utility in exploring centromere evolution and other repetitive genomic regions. Availability and implementation: Source code and example binaries freely available for download at github.com/gluncic/GRM2023.

It is known that the ~ 1.6 kb Neuroblastoma BreakPoint Family (NBPF) repeats are human specific and contributing to cognitive capabilities, with increasing frequency in higher order repeat 3mer HORs (Olduvai triplets). From chimpanzee to modern human there is a discontinuous jump from 0 to ~ 50 tandemly organized 3mer HORs. Here we investigate the structure of NBPF 3mer HORs in the Neanderthal genome assembly of Pääbo et al., comparing it to the results obtained for human hg38.p14 chromosome 1. Our findings reveal corresponding NBPF 3mer HOR arrays in Neanderthals with slightly different monomer structures and numbers of HOR copies compared to humans. Additionally, we compute the NBPF 3mer HOR pattern for the complete telomere-to-telomere human genome assembly (T2T-CHM13) by Miga et al., identifying two novel tandem arrays of NBPF 3mer HOR repeats with 5 and 9 NBPF 3mer HOR copies. We hypothesize that these arrays correspond to novel NBPF genes (here referred to as NBPFA1 and NBPFA2). Further improving the quality of the Neanderthal genome using T2T-CHM13 as a reference would be of great interest in determining the presence of such distant novel NBPF genes in the Neanderthal genome and enhancing our understanding of human evolution.

It is known that the ~ 1.6 kb NBPF repeats are human specific and contributing to cognitive capabilities, with increasing frequency in higher order repeat 3mer HORs (Olduvai triplets). From chimpanzee to modern human there is a discontinuous jump from 0 to ~ 50 tandemly organized 3mer HORs. Here we investigate the structure of NBPF 3mer HORs in the Neanderthal genome assembly of Pääbo and collaborators, comparing it to the results obtained for human hg38.p14 chromosome 1. Our findings reveal corresponding NBPF 3mer HOR arrays in Neanderthals with slightly different monomer structures and numbers of HOR copies compared to humans. Additionally, we compute the NBPF 3mer HOR pattern for the complete telomere-to-telomere human genome assembly (T2T-CHM13) by Miga and collaborators, identifying two novel tandem arrays of NBPF 3mer HOR repeats with 5 and 9 NBPF 3mer HOR copies. We hypothesize that these arrays correspond to novel NBPF genes (here referred to as NBPFA1 and NBPFA2). Further improving the quality of the Neanderthal genome using T2T-CHM13 as a reference would be of great interest in determining the presence of such distant novel NBPF genes in the Neanderthal genome and enhancing our understanding of human evolution.