Bernardo Aguzzoli Heberle
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Bernardo Aguzzoli Heberle has a total of 91 co-authors. Here are the top co-authors by number of shared publications:
- Mark T. W. Ebbert
Sanders-Brown Center On Aging and Alzheimer’s Disease Research Center, University of Kentucky, Lexington, USA
16 shared publications - Madeline L. Page
Sanders-Brown Center on Aging, University of Kentucky, Lexington, KY, USA
10 shared publications J. Anthony Brandon
Sanders-Brown Center on Aging, University of Kentucky, Lexington, USA
8 shared publications- David W. Fardo
University of Kentucky
6 shared publications Peter T. Nelson
Sanders-Brown Center On Aging and Alzheimer’s Disease Research Center, University of Kentucky, Lexington, USA
5 shared publications- Thiago Wendt Viola
Graduate Program in Medicine and Health Sciences, School of Medicine, Pontifical Catholic University of Rio Grande do Sul (PUCRS)
5 shared publications
Latest Publications
- Open Access
- 17 July 2026
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 54 (14)
Systematic contextual biases in SegmentNT potentially relevant to other nucleotide transformer models
Recent advances in large language models have extended to genomic applications, yet model robustness relative to context is unclear. Here, we demonstrate two intrinsic biases (input sequence length and nucleotide position) affecting SegmentNT results, a model included with the Nucleotide Transformer that provides nucleotide-level predictions of biological features. We demonstrate that nucleotide position within the input sequence (beginning, middle, or end) alters the nature of SegmentNT’s raw prediction probabilities, which can be standardized to improve prediction consistency. While longer input sequence length improves model performance, diminishing returns suggest a surprisingly small input length of ∼3072 nucleotides might be sufficient for many applications. We further identify a 24-nucleotide periodic oscillation in SegmentNT’s prediction probabilities, revealing an intrinsic bias potentially linked to the model’s training tokenization (6-mers) and architecture. We identify potential approaches to account for these biases and provide generalizable insights for utilizing nucleotide-resolution functional prediction models.
- 17 July 2026
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 54 (14)
Open AccessSystematic contextual biases in SegmentNT potentially relevant to other nucleotide transformer models
Recent advances in large language models have extended to genomic applications, yet model robustness relative to context is unclear. Here, we demonstrate two intrinsic biases (input sequence length and nucleotide position) affecting SegmentNT results, a model included with the Nucleotide Transformer that provides nucleotide-level predictions of biological features. We demonstrate that nucleotide position within the input sequence (beginning, middle, or end) alters the nature of SegmentNT’s raw prediction probabilities, which can be standardized to improve prediction consistency. While longer input sequence length improves model performance, diminishing returns suggest a surprisingly small input length of ∼3072 nucleotides might be sufficient for many applications. We further identify a 24-nucleotide periodic oscillation in SegmentNT’s prediction probabilities, revealing an intrinsic bias potentially linked to the model’s training tokenization (6-mers) and architecture. We identify potential approaches to account for these biases and provide generalizable insights for utilizing nucleotide-resolution functional prediction models.
- Open Access
- 5 December 2025
- Published by Springer Nature in BMC Genomics
- Vol. 26 (1) , 1078
A bioinformatic survey of RNA isoform diversity and expression across 9 GTEx tissues using long-read sequencing data
Background: Even though alternative RNA splicing was discovered nearly 50 years ago (1977), we still understand very little about most isoforms arising from a single gene, including in which tissues they are expressed and if their functions differ. Human gene annotations suggest remarkable transcriptional complexity, with approximately 252,798 distinct RNA isoform annotations from 62,710 gene bodies (Ensembl v109; 2023), emphasizing the need to understand their biological effects. For example, 256 gene bodies have ≥ 50 annotated isoforms, and 30 have ≥ 100, where one protein-coding gene ( MAPK10 ) even has 192 distinct RNA isoform annotations. Whether such isoform diversity results from biological redundancy or spurious alternative splicing (i.e., noise), or whether individual isoforms have specialized functions (even if subtle) remains a mystery for most genes. Three recent studies demonstrated that long-read RNAseq enables improved RNA isoform quantification for essentially any tissue, cell type, or biological condition (e.g., disease, development, aging, etc.), making it possible to better assess individual isoform expression and function. While each study provided important discoveries related to RNA isoform diversity, deeper exploration is needed. Results: We sought to quantify and characterize real isoform usage across tissues (compared to annotations). We used long-read RNAseq data from 58 GTEx samples across nine tissues (three brain, two heart, muscle, lung, liver, and cultured fibroblasts) generated by Glinos et al. and found considerable isoform diversity within and across tissues. Cerebellar hemisphere was the most transcriptionally complex tissue (22,522 distinct isoforms; 3,726 unique); liver was the least diverse (12,435 distinct isoforms; 1,039 unique). We highlight gene clusters exhibiting high tissue-specific isoform diversity per tissue (e.g., TPM1 expresses 19 in heart’s atrial appendage). We also validated 447 of the 700 new isoforms discovered by Aguzzoli-Heberle et al. and found that 88 were expressed in all nine tissues, while 58 were specific to a single tissue. Conclusions: This study represents a broad bioinformatic survey of the RNA isoform landscape, demonstrating isoform diversity across nine tissues and emphasizes the need for further verification, validation, and functional annotation research to better understand how individual isoforms from a single gene body contribute to human health and disease.
- 5 December 2025
- Published by Springer Nature in BMC Genomics
- Vol. 26 (1) , 1078
Open AccessA bioinformatic survey of RNA isoform diversity and expression across 9 GTEx tissues using long-read sequencing data
Background: Even though alternative RNA splicing was discovered nearly 50 years ago (1977), we still understand very little about most isoforms arising from a single gene, including in which tissues they are expressed and if their functions differ. Human gene annotations suggest remarkable transcriptional complexity, with approximately 252,798 distinct RNA isoform annotations from 62,710 gene bodies (Ensembl v109; 2023), emphasizing the need to understand their biological effects. For example, 256 gene bodies have ≥ 50 annotated isoforms, and 30 have ≥ 100, where one protein-coding gene ( MAPK10 ) even has 192 distinct RNA isoform annotations. Whether such isoform diversity results from biological redundancy or spurious alternative splicing (i.e., noise), or whether individual isoforms have specialized functions (even if subtle) remains a mystery for most genes. Three recent studies demonstrated that long-read RNAseq enables improved RNA isoform quantification for essentially any tissue, cell type, or biological condition (e.g., disease, development, aging, etc.), making it possible to better assess individual isoform expression and function. While each study provided important discoveries related to RNA isoform diversity, deeper exploration is needed. Results: We sought to quantify and characterize real isoform usage across tissues (compared to annotations). We used long-read RNAseq data from 58 GTEx samples across nine tissues (three brain, two heart, muscle, lung, liver, and cultured fibroblasts) generated by Glinos et al. and found considerable isoform diversity within and across tissues. Cerebellar hemisphere was the most transcriptionally complex tissue (22,522 distinct isoforms; 3,726 unique); liver was the least diverse (12,435 distinct isoforms; 1,039 unique). We highlight gene clusters exhibiting high tissue-specific isoform diversity per tissue (e.g., TPM1 expresses 19 in heart’s atrial appendage). We also validated 447 of the 700 new isoforms discovered by Aguzzoli-Heberle et al. and found that 88 were expressed in all nine tissues, while 58 were specific to a single tissue. Conclusions: This study represents a broad bioinformatic survey of the RNA isoform landscape, demonstrating isoform diversity across nine tissues and emphasizes the need for further verification, validation, and functional annotation research to better understand how individual isoforms from a single gene body contribute to human health and disease.
- Open Access
- 5 December 2025
- Published by Springer Nature in BMC Genomics
- Vol. 26 (1) , 1079
Genome annotations matter: characterizing Ensembl hg38 annotations from 2014 to 2023
Background: An accurate genome annotation is essential in many contexts, including RNA sequencing studies. Annotations include known genes and isoforms, detailing their location (chromosome, start, and end) and coding sequence, among other important metadata. Results: We characterized changes in human Ensembl annotations from 2014 to 2023 and the important gains in our biological understanding in recent years. While generally gene and isoform annotations increased (2014: 58,812 genes ; 2023: 62,710), some years dropped (e.g., 2016). A similar pattern exists for the gene and isoform biotypes; both 2015 (19,825) and 2017 (19,828) have fewer genes annotated as protein-coding than 2014 (19,953) and 2016 (19,961)— 2023 has the most (20,048). PCBP1-AS1 had the most annotated isoforms (296). We quantified expression for isoforms that were new between 2019 and 2023 across nine GTEx tissues (58 samples) to demonstrate our significant gains in understanding recently. We saw 2,054 of these ‘new’ isoforms expressed in cerebellar hemisphere (594 in liver). For many genes, we saw that the relative expression of the ‘new’ isoforms was much greater than the previously known isoforms. Conclusions: This study demonstrates the importance of an accurate genome annotation to truly understand the underlying complexity of biology that is often oversimplified by ignoring transcriptional complexity.
- 5 December 2025
- Published by Springer Nature in BMC Genomics
- Vol. 26 (1) , 1079
Open AccessGenome annotations matter: characterizing Ensembl hg38 annotations from 2014 to 2023
Background: An accurate genome annotation is essential in many contexts, including RNA sequencing studies. Annotations include known genes and isoforms, detailing their location (chromosome, start, and end) and coding sequence, among other important metadata. Results: We characterized changes in human Ensembl annotations from 2014 to 2023 and the important gains in our biological understanding in recent years. While generally gene and isoform annotations increased (2014: 58,812 genes ; 2023: 62,710), some years dropped (e.g., 2016). A similar pattern exists for the gene and isoform biotypes; both 2015 (19,825) and 2017 (19,828) have fewer genes annotated as protein-coding than 2014 (19,953) and 2016 (19,961)— 2023 has the most (20,048). PCBP1-AS1 had the most annotated isoforms (296). We quantified expression for isoforms that were new between 2019 and 2023 across nine GTEx tissues (58 samples) to demonstrate our significant gains in understanding recently. We saw 2,054 of these ‘new’ isoforms expressed in cerebellar hemisphere (594 in liver). For many genes, we saw that the relative expression of the ‘new’ isoforms was much greater than the previously known isoforms. Conclusions: This study demonstrates the importance of an accurate genome annotation to truly understand the underlying complexity of biology that is often oversimplified by ignoring transcriptional complexity.
- Open Access
Decoding the human PBMC isonome: Isoform-level resolution with single-cell long-read transcriptomics
Long-read single-cell RNA sequencing provides an opportunity to understand human health and disease at a level difficult to resolve with bulk or short-read methods. This approach enables isoform-level investigation of cellular diversity and disease mechanisms and definition of cell-types, rather than using genes alone. Using a modified, microfluidic-free PIPseq workflow and computational pipeline adapted for Oxford Nanopore long-read sequencing, we generated the largest long-read single-cell dataset of human peripheral blood mononuclear cells (PBMCs) to date. This study profiled isoform usage across immune cells, integrating marker expression and isoform discovery. We identified 128 novel isoforms from known and new genes, several with distinct cell-type-specific patterns, and characterized marker gene isoform expression across cell-types. Non-canonical protein-coding variants of GZMB and CD3G were enriched in unexpected cell-types, including megakaryocytes and monocyte-derived populations. We also discovered novel transcripts from CMC1 and LYAR with cell-type-specific signatures that were the predominantly expressed transcript within the gene. This study expands versatility of long-read single-cell studies to not only relay changes in isoform signatures, but to position them within the functional context of the biology they impact. These results demonstrate the power of long-read single-cell sequencing for mapping the isoform landscape—the isonome—across tissues and disease contexts.
Open AccessDecoding the human PBMC isonome: Isoform-level resolution with single-cell long-read transcriptomics
Long-read single-cell RNA sequencing provides an opportunity to understand human health and disease at a level difficult to resolve with bulk or short-read methods. This approach enables isoform-level investigation of cellular diversity and disease mechanisms and definition of cell-types, rather than using genes alone. Using a modified, microfluidic-free PIPseq workflow and computational pipeline adapted for Oxford Nanopore long-read sequencing, we generated the largest long-read single-cell dataset of human peripheral blood mononuclear cells (PBMCs) to date. This study profiled isoform usage across immune cells, integrating marker expression and isoform discovery. We identified 128 novel isoforms from known and new genes, several with distinct cell-type-specific patterns, and characterized marker gene isoform expression across cell-types. Non-canonical protein-coding variants of GZMB and CD3G were enriched in unexpected cell-types, including megakaryocytes and monocyte-derived populations. We also discovered novel transcripts from CMC1 and LYAR with cell-type-specific signatures that were the predominantly expressed transcript within the gene. This study expands versatility of long-read single-cell studies to not only relay changes in isoform signatures, but to position them within the functional context of the biology they impact. These results demonstrate the power of long-read single-cell sequencing for mapping the isoform landscape—the isonome—across tissues and disease contexts.
- Open Access
- 17 June 2025
- Published by Springer Nature in Nature Genetics
- Vol. 57 (7) , 1
Author Correction: GWAS of multiple neuropathology endophenotypes identifies new risk loci and provides insights into the genetic risk of dementia
- 17 June 2025
- Published by Springer Nature in Nature Genetics
- Vol. 57 (7) , 1
Open AccessAuthor Correction: GWAS of multiple neuropathology endophenotypes identifies new risk loci and provides insights into the genetic risk of dementia
- Open Access
- 17 June 2025
- Published by Springer Nature in Nature Genetics
- Vol. 57 (7) , 1791
Publisher Correction: GWAS of multiple neuropathology endophenotypes identifies new risk loci and provides insights into the genetic risk of dementia
- 17 June 2025
- Published by Springer Nature in Nature Genetics
- Vol. 57 (7) , 1791
Open AccessPublisher Correction: GWAS of multiple neuropathology endophenotypes identifies new risk loci and provides insights into the genetic risk of dementia
- Open Access
Sequencing the gaps: dark genomic regions persist in CHM13 despite long-read advances
Comprehensive genomic analysis is essential for advancing our understanding of human genetics and disease. However, short-read sequencing technologies are inherently limited in their ability to resolve highly repetitive, structurally complex, and low-mappability genomic regions, previously coined as “dark” regions. Long-read sequencing technologies, such as PacBio and Oxford Nanopore Technologies (ONT), offer improved resolution of these regions, yet they are not perfect. With the advent of the new Telomere-to-Telomere (T2T) CHM13 reference genome, exploring its effect on dark regions is prudent. In this study, we systematically analyze dark regions across four human genome references—HG19, HG38 (with and without alternate contigs), and CHM13—using both short- and long-read sequencing data. We found that dark regions increase as the reference becomes more complete, especially dark-by-MAPQ regions, but that long-read sequencing significantly reduces the number of dark regions in the genome, particularly within gene bodies. However, we identify potential alignment challenges in long-read data, such as centromeric regions. These findings highlight the importance of both reference genome selection and sequencing technology choice in achieving a truly comprehensive genomic analysis.
Open AccessSequencing the gaps: dark genomic regions persist in CHM13 despite long-read advances
Comprehensive genomic analysis is essential for advancing our understanding of human genetics and disease. However, short-read sequencing technologies are inherently limited in their ability to resolve highly repetitive, structurally complex, and low-mappability genomic regions, previously coined as “dark” regions. Long-read sequencing technologies, such as PacBio and Oxford Nanopore Technologies (ONT), offer improved resolution of these regions, yet they are not perfect. With the advent of the new Telomere-to-Telomere (T2T) CHM13 reference genome, exploring its effect on dark regions is prudent. In this study, we systematically analyze dark regions across four human genome references—HG19, HG38 (with and without alternate contigs), and CHM13—using both short- and long-read sequencing data. We found that dark regions increase as the reference becomes more complete, especially dark-by-MAPQ regions, but that long-read sequencing significantly reduces the number of dark regions in the genome, particularly within gene bodies. However, we identify potential alignment challenges in long-read data, such as centromeric regions. These findings highlight the importance of both reference genome selection and sequencing technology choice in achieving a truly comprehensive genomic analysis.
- Open Access
Systematic contextual biases in SegmentNT potentially relevant to other nucleotide transformer models
Recent advances in large language models (LLMs) have extended to genomic applications, yet model robustness relative to context is unclear. Here, we demonstrate two intrinsic biases (input sequence length and nucleotide position) affecting SegmentNT results, a model included with the Nucleotide Transformer that provides nucleotide-level predictions of biological features. We demonstrate that nucleotide position within the input sequence (beginning, middle, or end) alters the nature of SegmentNT’s raw prediction probabilities, which can be standardized to improve prediction consistency. While longer input sequence length improves model performance, diminishing returns suggest a surprisingly small input length of ∼3,072 nucleotides might be sufficient for many applications. We further identify a 24-nucleotide periodic oscillation in SegmentNT’s prediction probabilities, revealing an intrinsic bias potentially linked to the model’s training tokenization (6-mers) and architecture. We identify potential approaches to account for these biases and provide generalizable insights for utilizing nucleotide-resolution functional prediction models.
Open AccessSystematic contextual biases in SegmentNT potentially relevant to other nucleotide transformer models
Recent advances in large language models (LLMs) have extended to genomic applications, yet model robustness relative to context is unclear. Here, we demonstrate two intrinsic biases (input sequence length and nucleotide position) affecting SegmentNT results, a model included with the Nucleotide Transformer that provides nucleotide-level predictions of biological features. We demonstrate that nucleotide position within the input sequence (beginning, middle, or end) alters the nature of SegmentNT’s raw prediction probabilities, which can be standardized to improve prediction consistency. While longer input sequence length improves model performance, diminishing returns suggest a surprisingly small input length of ∼3,072 nucleotides might be sufficient for many applications. We further identify a 24-nucleotide periodic oscillation in SegmentNT’s prediction probabilities, revealing an intrinsic bias potentially linked to the model’s training tokenization (6-mers) and architecture. We identify potential approaches to account for these biases and provide generalizable insights for utilizing nucleotide-resolution functional prediction models.
- Open Access
- 5 March 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (3) , e70025
Systematic review and meta‐analysis of bulk RNAseq studies in human Alzheimer's disease brain tissue
We systematically reviewed and meta‐analyzed bulk RNA sequencing (RNAseq) studies comparing Alzheimer's disease (AD) patients to controls in human brain tissue. We searched PubMed, Web of Science, and Scopus for human brain bulk RNAseq studies, excluding re‐analyses and studies limited to small RNAs or gene panels. We developed 10 criteria for quality assessment and performed a meta‐analysis on three high‐quality datasets. Of 3266 records, 24 qualified for the systematic review, and one study with three datasets qualified for the meta‐analysis. The meta‐analysis identified 571 differentially expressed genes (DEGs) in the temporal lobe and 189 in the frontal lobe, including CLU and GFAP. Pathway analysis suggested reactivation of developmental processes in the adult AD brain. Limited data availability constrained the meta‐analysis. These findings underscore the need for rigorous methods in AD transcriptomic research to better identify transcriptomic changes and advance biomarker and therapeutic development. This review is registered in PROSPERO (CRD42023466522).Highlights: Comprehensive review: Conducted the first systematic review and meta‐analysis of bulk RNA sequencing (RNAseq) studies comparing Alzheimer's disease (AD) patients with non‐demented controls using primary human brain tissue. Key findings: Identified 571 differentially expressed genes (DEGs) in the temporal lobe and 189 in the frontal lobe of patients with AD, revealing potential therapeutic targets. Pathway discovery: Highlighted key overlapping pathways such as “tube morphogenesis” and “neuroactive ligand‐receptor interaction” that may play critical roles in AD. Quality assessment: Emphasized the importance of methodological rigor in transcriptomic studies, including quality assessment tools to guide future research in AD. Study limitation: Acknowledged limited access to complete data tables and lack of diversity in existing datasets, which constrained some of the analysis.
- 5 March 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (3) , e70025
Open AccessSystematic review and meta‐analysis of bulk RNAseq studies in human Alzheimer's disease brain tissue
We systematically reviewed and meta‐analyzed bulk RNA sequencing (RNAseq) studies comparing Alzheimer's disease (AD) patients to controls in human brain tissue. We searched PubMed, Web of Science, and Scopus for human brain bulk RNAseq studies, excluding re‐analyses and studies limited to small RNAs or gene panels. We developed 10 criteria for quality assessment and performed a meta‐analysis on three high‐quality datasets. Of 3266 records, 24 qualified for the systematic review, and one study with three datasets qualified for the meta‐analysis. The meta‐analysis identified 571 differentially expressed genes (DEGs) in the temporal lobe and 189 in the frontal lobe, including CLU and GFAP. Pathway analysis suggested reactivation of developmental processes in the adult AD brain. Limited data availability constrained the meta‐analysis. These findings underscore the need for rigorous methods in AD transcriptomic research to better identify transcriptomic changes and advance biomarker and therapeutic development. This review is registered in PROSPERO (CRD42023466522).Highlights: Comprehensive review: Conducted the first systematic review and meta‐analysis of bulk RNA sequencing (RNAseq) studies comparing Alzheimer's disease (AD) patients with non‐demented controls using primary human brain tissue. Key findings: Identified 571 differentially expressed genes (DEGs) in the temporal lobe and 189 in the frontal lobe of patients with AD, revealing potential therapeutic targets. Pathway discovery: Highlighted key overlapping pathways such as “tube morphogenesis” and “neuroactive ligand‐receptor interaction” that may play critical roles in AD. Quality assessment: Emphasized the importance of methodological rigor in transcriptomic studies, including quality assessment tools to guide future research in AD. Study limitation: Acknowledged limited access to complete data tables and lack of diversity in existing datasets, which constrained some of the analysis.
- Open Access
- 26 December 2024
- Published by Oxford University Press (OUP) in Bioinformatics Advances
- Vol. 5 (1) , vbaf057
RNApysoforms: fast rendering interactive visualization of RNA isoform structure and expression in Python
Summary: Alternative splicing generates multiple RNA isoforms from a single gene, enriching genetic diversity and impacting gene function. Effective visualization of these isoforms and their expression patterns is crucial but challenging due to limitations in existing tools. Traditional genome browsers lack programmability, while other tools offer limited customization, produce static plots, or cannot simultaneously display structures and expression levels. RNApysoforms was developed to address these gaps by providing a Python-based package that enables concurrent visualization of RNA isoform structures and expression data. Leveraging plotly and polars libraries, it offers an interactive, customizable, and faster-rendering framework suitable for web applications, enhancing the analysis and dissemination of RNA isoform research. Availability and implementation: RNApysoforms is a Python package available at (https://github.com/UK-SBCoA-EbbertLab/RNApysoforms) and (https://zenodo.org/records/14941190) via an open-source MIT license. It can be easily installed using the pip package installer for Python. Thorough documentation and usage vignettes are available at: https://rna-pysoforms.readthedocs.io/en/latest/.
- 26 December 2024
- Published by Oxford University Press (OUP) in Bioinformatics Advances
- Vol. 5 (1) , vbaf057
Open AccessRNApysoforms: fast rendering interactive visualization of RNA isoform structure and expression in Python
Summary: Alternative splicing generates multiple RNA isoforms from a single gene, enriching genetic diversity and impacting gene function. Effective visualization of these isoforms and their expression patterns is crucial but challenging due to limitations in existing tools. Traditional genome browsers lack programmability, while other tools offer limited customization, produce static plots, or cannot simultaneously display structures and expression levels. RNApysoforms was developed to address these gaps by providing a Python-based package that enables concurrent visualization of RNA isoform structures and expression data. Leveraging plotly and polars libraries, it offers an interactive, customizable, and faster-rendering framework suitable for web applications, enhancing the analysis and dissemination of RNA isoform research. Availability and implementation: RNApysoforms is a Python package available at (https://github.com/UK-SBCoA-EbbertLab/RNApysoforms) and (https://zenodo.org/records/14941190) via an open-source MIT license. It can be easily installed using the pip package installer for Python. Thorough documentation and usage vignettes are available at: https://rna-pysoforms.readthedocs.io/en/latest/.