Tuuli Lappalainen

To evaluate the clinical implementation of long-read genome sequencing (lrGS) and establish the population reference resources needed to support its use in diagnostics, we integrate two complementary initiatives: Genome of Sweden (GoS), Sweden’s contribution to Genome of Europe (GoE), and the Genomic Medicine Sweden – Rare Diseases (GMS-RD) long-read sequencing pilot. The initial GoS release of 576 individuals provides population-scale maps of genetic and epigenetic variation. Among 485 GMS-RD probands, 161 (33.2%) receive a molecular diagnosis, with lrGS providing added clinical value in 40 (24.8%). As a first-line test, lrGS could have avoided at least 42 follow-up analyses and an additional testing cycle in 7.2% of probands, reducing costs and the burden on patients. By integrating population and clinical long-read genomics, Sweden establishes a national ecosystem for genomic research and clinical diagnostics. Our findings support the clinical implementation of lrGS and offer a scalable model for national collaboration.

Single-cell CRISPR screens have enabled systematic investigation of gene function, but studies have largely focused on gene-level effects, overlooking transcriptional complexity and isoform usage. Methods capable of capturing splicing and isoform usage have emerged, including long-read sequencing and alternative library preparation strategies, but their suitability for large-scale perturbation screens remains unevaluated. We compare two library preparation methods (10x Genomics and Parse Biosciences) across Illumina short-read, Oxford Nanopore, and PacBio long-read sequencing, applying CRISPRi to silence three genes with distinct regulatory roles (DDX6, GEMIN5, GFI1B) in K562 cells. While short-read methods detected some splicing events, only long-read sequencing consistently captured isoform-level changes. Although Parse provided even transcript coverage, we observed strong intronic read enrichment, limiting its utility for splicing analysis. The primary constraint of long-read approaches was sequencing depth: ~21 million reads are needed for 80% saturation of splicing events in a single perturbation. Notably, GEMIN5 knockdown produced only modest differential expression but the most extensive splicing changes, an effect invisible to gene-level analysis, underscoring the value of isoform-level screens. We provide a practical framework for isoform-level analysis in single-cell CRISPR screens, identifying current capabilities and limitations. As perturbation studies scale, long-read sequencing will be essential for comprehensive functional interpretation, capturing biology missed by gene-level analysis.

Over the past two decades, genome-wide association studies (GWAS) have identified thousands of trait- and disease-associated loci. However, the mechanistic understanding of these loci remains incomplete, which limits our ability to understand gene regulation and cellular programs underlying complex traits, predict disease risk, and develop therapeutics targeted to root causes. Here, we describe the current challenges for using GWAS to prioritize variants for functional follow-up experiments. These challenges span multiple domains, including limitations in data sharing and harmonization, limitations of statistical and functional fine-mapping, and the ambiguity in the added value of emerging deep learning frameworks for variant effect prediction as a complementary approach alongside traditional statistical genetics methods. We analyze these variant prioritization methods and suggest a multi-modal approach for resolving GWAS loci to a focused set of high-confidence variants for functional exploration. Fully realizing the potential of GWAS will require harmonized summary statistics and broader sharing of in-sample linkage disequilibrium (LD) data to enable robust and scalable causal variant prioritization.

The impact of artificial intelligence (AI) on healthcare has lagged behind other sectors. In particular, the use of large-scale health data for AI-driven discovery of disease mechanisms, predict detailed outcomes, and guide precision medicine remains limited. Many of these unrealised goals can be achieved in the Nordic countries, which harbor the only large-scale, longitudinal, multimodal data sets currently available to produce generisable models for responsible, AI-based discovery in medicine, compatible with EHDS. We here present the Nordic AI-Health initiative, outline the rationale and infrastructure, and describe a roadmap for building a world-leading infrastructure with transformative impact on global healthcare.

Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease with substantial genetic and clinical heterogeneity that impedes therapeutic development. Large-scale multi-tissue genomic resources have transformed the study of neuropsychiatric and neurodegenerative diseases, but no equivalent resource exists for ALS. Here we present the full NYGC ALS Consortium dataset, combining whole-genome sequencing from 4,746 donors and bulk RNA-seq from 2,574 samples across 8 brain and spinal cord regions from 695 donors across the ALS disease spectrum. Our catalogue of small variants, structural variants, and short tandem repeats identified likely pathogenic mutations in 15.6% of ALS cases. Gene expression and mRNA splicing analysis across 5 major tissues reveals shared and region-specific features, highlighting microglial and T-cell dysregulation in the spinal cord. Mapping the genetic regulation of expression and splicing across tissues identified associations with 6 ALS risk loci, whereas allele-specific rare variant analysis detected expression effects for C9orf72 and OPTN . All data are immediately publicly available.

Osteoporosis is characterized by low bone mineral density (BMD) and elevated fracture risk. Most BMD-associated GWAS variants lie in noncoding regions, complicating efforts to identify causal genes and mechanisms. To overcome this variant-to-function challenge, we previously developed STING-seq, a framework integrating biobank-scale GWAS with single-cell CRISPR inhibition (CRISPRi) screening to directly connect noncoding cis -regulatory elements (CREs) to their target genes. Applied to human fetal osteoblast (hFOB) cells across osteogenic differentiation, STING-seq linked 76 CREs to 75 target genes at BMD loci. Arrayed CRISPRi and activation validated key CRE–gene relationships, including long-range enhancers regulating CXCL12 (over 500 kb apart). We further uncovered trans -regulatory networks and characterized the DAP3 – YY1AP1 bidirectional promoter, demonstrating a role for DAP3 in mineralization via mitochondrial pathways. Together, these findings provide mechanistic insight into how noncoding GWAS variants shape osteoblast activity and highlight the genes and pathways that mediate genetic effects on BMD.

Functional interpretation is essential for understanding how genetic variants contribute to complex traits. Here, we identified and characterized regulatory variants in CD4+ T cells collected from 362 donors. We integrated molecular QTL mapping from single-cell RNA-seq profiles and chromatin accessibility with predicted variant effects from a deep learning model trained on chromatin accessibility data. We identified molecular features and transcription factor binding mechanisms underlying variant sharing and mediated effects across the modalities and approaches. While predicted variant effects correlated with molQTLs, only a small fraction of empirically detected molQTLs were discovered by predictive models. MolQTLs, primarily those affecting chromatin, indicated potential molecular drivers for 33% of immune-related GWAS loci, with the deep learning approach providing insights into 4.7% of GWAS loci. These results highlight the value of multi-omic data and systematic integration of empirical and predictive approaches to interpret regulatory effects of genetic variants.

Genomic loci associated with common traits and diseases are typically non-coding and likely impact gene expression, sometimes coinciding with rare loss-of-function variants in the target gene. However, our understanding of how gradual changes in gene dosage affect molecular, cellular, and organismal traits is currently limited. To address this gap, we induced gradual changes in gene expression of four genes using CRISPR activation and inactivation in human-derived K562 cells. Downstream transcriptional consequences of dosage modulation of three master trans-regulators associated with blood cell traits ( GFI1B , NFE2 , and MYB ) were examined using targeted single-cell multimodal sequencing. We showed that guide tiling around the TSS is the most effective way to modulate cis gene expression across a wide range of fold changes, with further effects from chromatin accessibility and histone marks that differ between the inhibition and activation systems. Our single-cell data allowed us to precisely detect subtle to large gene expression changes in dozens of trans genes, revealing that many responses to dosage changes of these three TFs are nonlinear, including non-monotonic behaviours, even when constraining the fold changes of the master regulators to a copy number gain or loss. We found that the dosage properties are linked to gene constraint and that some of these nonlinear responses are enriched for disease and GWAS genes. Overall, our study provides a straightforward and scalable method to precisely modulate gene expression and gain insights into its downstream consequences at high resolution.

Allele-specific expression (ASE) outlier detection is a powerful tool for identifying genes affected by large effect rare genetic regulatory variants but suffers from data sparsity and noisy signal in low-count genes. Genome phasing can be utilized to aggregate ASE signal along haplotypes to alleviate both sparsity and noise. Yet statistical tools for utilizing haplotype-level ASE data for rare variant interpretation are lacking. Here, we present ANEVA-h, to quantify the amount of genetic variation in gene expression from haplotype-level ASE data in a population, enabling more accurate and comprehensive detection of regulatory effects. We apply ANEVA-h to GTEx project data, along with a compatible dosage outlier test, to show an over 2-fold increase in the number of testable genes, reduction of spurious outlier calls, and improved enrichment for rare high-impact variants. In clinical cohorts of neuromuscular and congenital heart disease, it enhances gene prioritization and identifies candidate diagnoses missed by DROP-MAE and ANEVA. Finally, we analyze globally diverse populations to characterize the impact of ancestry background in reference and the test population. We provide tools and data necessary to facilitate integration of haplotype level ASE outlier testing in rare variant interpretation pipelines

We analyze 10,986 participants (mean age 77; 63% women; 54% non-White) across seven U.S. cohorts to study the relationship between mitochondrial DNA (mtDNA) heteroplasmy and nuclear DNA methylation. We identify 597 CpGs associated with heteroplasmy burden, generally showing lower methylation. These CpGs are enriched in dynamically regulated island shores and depleted in CpG islands, indicating involvement in context-specific rather than constitutive gene regulation. In HEK293T cells, we introduce a truncating mtDNA mutation (MT-COX3, mt.9979) and observe a positive correlation between variant allele fraction and methylation at cg04569152, supporting a direct mtDNA–nDNA epigenetic link. Many heteroplasmy-associated CpGs overlap with known methylation-trait associations for metabolic and behavioral traits. Composite CpG scores predict all-cause mortality and incident CVD, with one-unit increases associated with 1.27-fold and 1.12-fold higher hazards, respectively. These findings suggest an mtDNA–nDNA epigenetic connection in aging and disease, though its direction and mechanisms remain to be studied.