Circulating tumor cells are the key link between a primary tumor and distant metastases, but once in the bloodstream, loss of adhesion induces cell death. To identify the mechanisms relevant for melanoma circulating tumor cell survival, we performed RNA sequencing and discovered that detached melanoma cells and isolated melanoma circulating tumor cells rewire lipid metabolism by upregulating fatty acid (FA) transport and FA beta-oxidation‒related genes. In patients with melanoma, high expression of FA transporters and FA beta-oxidation enzymes significantly correlates with reduced progression-free and overall survival. Among the highest expressed regulators in melanoma circulating tumor cells were the carnitine transferases carnitine O-octanoyltransferase and carnitine acetyltransferase, which control the shuttle of peroxisome-derived medium-chain FAs toward mitochondria to fuel mitochondrial FA beta-oxidation. Knockdown of carnitine O-octanoyltransferase or carnitine acetyltransferase and short-term treatment with peroxisomal or mitochondrial FA beta-oxidation inhibitors thioridazine or ranolazine suppressed melanoma metastasis in mice. Carnitine O-octanoyltransferase and carnitine acetyltransferase depletion could be rescued by medium-chain FA supplementation, indicating that the peroxisomal supply of FAs is crucial for the survival of nonadherent melanoma cells. Our study identifies targeting the FA-based cross-talk between peroxisomes and mitochondria as a potential therapeutic opportunity to challenge melanoma progression. Moreover, the discovery of the antimetastatic activity of the Food and Drug Administration‒approved drug ranolazine carries translational potential.

With the emergence of single-cell foundation models, an important question arises: how do these models perform when trained on datasets having an imbalance in cell type distribution due to rare cell types or biased sampling? We benchmark three foundation models, scGPT, scBERT, and Geneformer, using skewed single-cell cell-type distribution for cell-type annotation. While all models had reduced performance when challenged with rare cell types, scGPT and scBERT, performed better than Geneformer. Notably, in contrast to scGPT and scBERT, Geneformer uses ordinal positions of the tokenized genes rather than actual raw gene expression values. To mitigate the effect of a skewed distribution, we find that random oversampling, but not random undersampling, improved the performance for all three foundation models. Finally, scGPT, using FlashAttention, has the fastest computational speed, whereas scBERT is more memory-efficient. We conclude that tokenization and data representation are essential areas of research, and new strategies are needed to mitigate the effects of imbalanced learning in single-cell foundation models. Code and data for reproducibility are available at https://github.com/SabbaghCodes/ImbalancedLearningForSingleCellFoundationModels .

Dysregulation of early B cell lymphopoiesis—the process guiding cellular immunity development—can lead to malignancy, making it crucial to understand its regulatory mechanisms. We generated a multiomics resource comprising paired chromatin accessibility and gene expression profiles across eight human B cell precursor populations, providing a detailed characterization of early human B cell development. Integrative analysis revealed highly cell type–specific regulatory elements and enabled the reconstruction of the gene regulatory network governing differentiation. We identified putative candidate regulons, such as ELK3, enriched in pro–B cells and potentially involved in cell cycle progression. Regulons from bulk data were projected onto single-cell data, validating their activity and refining the regulatory landscape. This resource enabled identification of active regulatory programs and transformation-associated states in B cell acute lymphoblastic leukemia. The publicly available atlas provides a valuable resource for understanding B cell development and disease, supporting future efforts to decode regulatory programs in immunity and hematologic malignancies.

Large-scale genomic studies have significantly increased our knowledge of genetic variability across populations. Regional genetic profiling is essential for distinguishing common benign variants from disease-causing ones. To this end, we conducted a comprehensive characterization of exonic variants in the population of Navarre (Spain), utilizing whole genome sequencing data from 358 unrelated individuals of Spanish origin. Our analysis revealed 61,410 biallelic single nucleotide variants (SNV) within the Navarrese cohort, with 35% classified as common (MAF > 1%). By comparing allele frequency data from 1000 Genome Project (excluding the Iberian cohort of Spain, IBS), Genome Aggregation Database, and a Spanish cohort (including IBS individuals and data from Medical Genome Project), we identified 1069 SNVs common in Navarre but rare (MAF ≤ 1%) in all other populations. We further corroborated this observation with a second regional cohort of 239 unrelated exomes, which confirmed 676 of the 1069 SNVs as common in Navarre. In conclusion, this study highlights the importance of population-specific characterization of genetic variation to improve allele frequency filtering in sequencing data analysis to identify disease-causing variants.

Purpose: Large-scale genomic studies have significantly increased our knowledge of genetic variability across populations. Regional genetic profiling is essential for distinguishing common benign variants from disease-causing ones. To this end, we conducted a comprehensive characterization of exonic variants in the population of Navarre (Spain). Methods: Genome sequencing data from 358 unrelated individuals of Spanish origin from the Navarrese population were used. Results: Our analysis revealed 61,410 biallelic exonic single nucleotide variants (SNV) within the Navarrese cohort, with 35% classified as common (MAF > 1%). By comparing allele frequency data from 1000 Genome Project (excluding the Iberian cohort of Spain, IBS), Genome Aggregation Database, and a Spanish cohort (including IBS individuals and data from Medical Genome Project), we identified 1,069 SNVs common in Navarre but rare (MAF ≤ 1%) in all other populations. We further corroborated this observation with a second regional cohort of 239 unrelated exomes, which confirmed 676 of the 1,069 SNVs as common in Navarre. Conclusion: This study highlights the importance of population-specific characterization of genetic variation to improve allele frequency filtering in sequencing data analysis to identify disease-causing variants.