Simon Rasmussen

multiomeR provides a reproducible R pipeline for analyzing large-scale single-nucleus Epi Multiome datasets, improving efficiency and reducing memory usage compared to existing workflows.

The paper applies variational autoencoders to integrate biomedical data, explores model dynamics, and evaluates methods using synthetic and real datasets.

The paper evaluates Olink Explore plasma protein assays by analyzing genetic associations, cross-platform concordance, and tissue atlases to identify targets with low variance explainability.

SLIM predicts single-cell genetic perturbation responses using STRING embeddings and a closed-form ridge regression model, achieving competitive accuracy with low computational cost.

This paper introduces DoBSeq, a low-cost framework for detecting rare variants in population-scale genetic screening, achieving 95% variant detection and recovering all clinically relevant pathogenic variants in ACMG genes.

The paper constructs disease maps by integrating omics data with protein network embeddings, identifying more disease-relevant proteins than omics analysis alone.

TaxVAMB improves metagenome binning by combining tetranucleotide frequencies and co-abundances with taxonomic annotations, outperforming existing methods on benchmark and real datasets.

The paper introduces CLASTER, a deep learning model that predicts nascent transcription levels from chromatin landscape and 3D structure data, achieving high correlation with experimental measurements.

PlasMAAG improves plasmid reconstruction from metagenomics data using assembly-alignment graphs and contrastive learning, outperforming existing methods in accuracy and efficiency.

DoBSeqWF is a pipeline for analyzing pooled sequencing data to detect rare genetic variants, improving sensitivity with machine learning filters.

The paper applies a variational autoencoder to multi-omics data from the UK Biobank to assess disease risk prediction and the impact of preprocessing on model performance.

This study maps the proteomic changes in plasma, saliva, and urine after acute exercise, identifying over 3,000 regulated proteins and their associations with disease and drug targets.

This study identifies genetic regulation of plasma proteins in children and adolescents, finding obesity- and age-dependent pQTLs and potential causal links to cardiometabolic traits.

EIRsurvival predicts time-to-diagnosis using genetic and multi-omics data from large cohorts, achieving C-indices between 0.55 and 0.96.

This study identifies genetic determinants of plasma protein levels in children and adolescents using MS-based proteomics, revealing extensive genetic impacts that persist into adulthood.

This paper introduces a haplotype-based framework for heritability estimation and reference-free polygenic prediction, showing improved performance over genotype-based methods in depression and schizophrenia.

Taxometer improves taxonomic classification of metagenomics contigs by combining abundance profiles and tetra-nucleotide frequencies, enhancing accuracy across multiple tools.

The paper identifies non-linear genetic regulation of plasma proteins using a deep learning method, finding dominance and epistasis effects in large cohorts.

This paper introduces a deep learning method for imputing missing values in label-free quantitative proteomics data and shows its effectiveness in identifying differentially abundant proteins in alcohol-related liver disease.

BinBencher provides a benchmarking suite for metagenomic binning that produces more biologically meaningful evaluations than existing methods.

FAVA infers high-quality functional protein interaction networks from scRNA-seq and proteomics data, outperforming existing methods and identifying interactions for understudied proteins.

This study identifies an N-terminal signal peptide that enables efficient loading of therapeutics into the Afp nano-injection system for targeted delivery.

This paper introduces AAMB, a deep learning method for metagenomic binning that improves genome reconstruction by integrating sequence co-abundances and tetranucleotide frequencies, outperforming existing methods in completeness and diversity.

The paper traces the evolutionary history of the CCR5delta32 deletion using ancient and modern genomes, finding it originated at least 7,000 years ago in Western Eurasia and underwent positive selection.

The paper introduces a deep learning framework for polygenic risk score prediction, showing improved performance over traditional methods for various diseases.

This study uses deep learning to predict blood and urine biomarkers from genetic data, outperforming linear methods and capturing non-linear and epistatic effects.

The paper evaluates metagenomic software tools using realistic datasets, highlighting improvements in assembly and taxon profiling while identifying remaining challenges in virus/archaea detection and reproducibility.

This study identifies 10% of pediatric CNS tumors as being caused by rare germline variants in known cancer predisposition genes through whole-genome sequencing and evolutionary constraint analysis.

The paper develops PHAMB, a method to bin viral genomes from bulk metagenomics data, and applies it to the HMP2 cohort to recover viral genomes and explore viral-host interactions.

This paper introduces a deep learning method using variational autoencoders to improve microbial genome binning from metagenomic data, achieving higher precision and completeness in reconstructed bins.