Anna Minkina

The mitochondrial genome (mtDNA), rich in repeats and prone to nuclear mitochondrial DNA segments (NUMTs), is implicated in cancer, metabolic syndromes, and neurodegeneration, yet short-read sequencing yields incomplete catalogs, mapping artifacts, and false heteroplasmies. We introduce MitoScope, a scalable end-to-end long-read mitochondrial analysis workflow that integrates mtDNA-read selection, heteroplasmic variant detection, assembly, copy number estimation, read-level haplotyping, and NUMT profiling. Applied to benchmarking data from the Somatic Mosaicism Across Human Tissues (SMaHT) Network, we reveal high sensitivity and precision compared to existing methods and successfully uncover known mitochondrial biology relevant for variant interpretation, aging, and human disease.

Oncogenesis involves widespread genetic and epigenetic alterations, yet the full spectrum of somatic variation genome-wide remains unresolved. We generated a near-telomere-to-telomere (T2T) diploid assembly of a donor paired with deep short- and long-read sequencing of their melanoma. This revealed that 16% of somatic variants occur in sequences absent from GRCh38, with satellite repeats acting as hotspots for UV-induced damage due to sequence-intrinsic mutability and inefficient repair. Centromere kinetochore domains emerged as focal sites of structural, genetic, and epigenetic variation, leading to remodeling of centromere kinetochore binding domains during tumor evolution. Single-molecule telomere reconstructions uncovered cycles of attrition, deletion, and telomerase-mediated extension that shape cancer telomeres. Finally, diploid chromatin maps exposed that copy number alterations and epimutations, rather than point mutations, predominate in rewiring cancer regulatory programs. These findings define the full landscape of a cancer’s somatic variation and their functional impact, establishing a blueprint for T2T studies of mosaicism.

Aberrant expression of stem-cell-associated genes is a common feature in acute myeloid leukemia (AML) and is linked to leukemic self-renewal and therapy resistance. Using AF10-rearranged leukemia as a prototypical example displaying a recurrent “ stemness ” network activated in AML, we screened for chromatin regulators that sustain aberrant activation of these networks. We deployed a CRISPR-Cas9 screen with a bespoke domain-focused library and identified several novel chromatin-modifying complexes as regulators of the TALE domain transcription factor MEIS1, a key leukemia stem cell (LSC)-associated gene. CRISPR droplet sequencing revealed that many of these MEIS1 regulators coordinately controlled the transcription of several AML oncogenes. In particular, we identified a novel role for the Tudor-domain containing chromatin reader protein SGF29 in the transcription of key AML oncogenes. Furthermore, SGF29 deletion impaired leukemogenesis in models representative of multiple AML subtypes. Our studies reveal a novel role for SGF29 as a non-oncogenic dependency in AML and identify the SGF29 Tudor domain as an attractive target for drug discovery.

Gene expression heterogeneity is ubiquitous within single cell datasets, even among cells of the same type. Heritable expression differences, defined here as those which persist over multiple cell divisions, are of particular interest, as they can underlie processes including cell differentiation during development as well as the clonal selection of drug-resistant cancer cells. However, heritable sources of variation are difficult to disentangle from non-heritable ones, such as cell cycle stage, asynchronous transcription, and measurement noise. Since heritable states should be shared by lineally related cells, we sought to leverage CRISPR-based lineage tracing, together with single cell molecular profiling, to discriminate between heritable and non-heritable variation in gene expression. We show that high efficiency capture of lineage profiles alongside single cell gene expression enables accurate lineage tree reconstruction and reveals an abundance of progressive, heritable gene expression changes. We find that a subset of these are likely mediated by structural genetic variation (copy number alterations, translocations), but that the stable attributes of others cannot be understood with expression data alone. Towards addressing this, we develop a method to capture cell lineage histories alongside single cell chromatin accessibility profiles, such that expression and chromatin accessibility of closely related cells can be linked via their lineage histories. We call this indirect “coassay” approach “THE LORAX” and leverage it to explore the genetic and epigenetic mechanisms underlying heritable gene expression changes. Using this approach, we show that we can discern between heritable gene expression differences mediated by large and small copy number changes, trans effects, and possible epigenetic variation.

Over a million DNA regulatory elements have been cataloged in the human genome, but linking these elements to the genes that they regulate remains challenging. We introduce Cicero, a statistical method that connects regulatory elements to target genes using single cell chromatin accessibility data. We apply Cicero to investigate how thousands of dynamically accessible elements orchestrate gene regulation in differentiating myoblasts. Groups of co-accessible regulatory elements linked by Cicero meet criteria of “chromatin hubs”, in that they are physically proximal, interact with a common set of transcription factors, and undergo coordinated changes in histone marks that are predictive of gene expression. Pseudotemporal analysis revealed a subset of elements bound by MYOD in myoblasts that exhibit early opening, potentially serving as the initial sites of recruitment of chromatin remodeling and histone-modifying enzymes. The methodological framework described here constitutes a powerful new approach for elucidating the architecture, grammar and mechanisms of cis- regulation on a genome-wide basis.