Esa Pitkänen

Towards a foundation model for raw nanopore signal through discrete tokenization

This paper introduces Brieno, a self-supervised model for raw nanopore signals that improves read classification and methylation detection without reference genomes.

petVAE: A Data-Driven Model for Identifying Amyloid PET Subgroups Across the Alzheimer's Disease Continuum

petVAE identifies biologically meaningful subgroups along the Alzheimer's disease continuum using a deep learning model trained on amyloid PET scans.

muat: portable transformer-based method for tumour classification and representation learning from somatic variants

The paper introduces muat, a transformer-based tool for tumour classification using somatic variants, achieving high accuracy across different cohorts without retraining.

Pan-cancer tumour classification and risk stratification from whole-genome somatic variants via dual-task representation learning

MuAt2, a Transformer model, classifies tumour types and subtypes from somatic variants, improving accuracy and calibration across cancer types and cohorts.

Classical enhancers couple cis-regulatory logic with transcriptional condensates and 3D genome architecture

The paper shows classical enhancers influence 3D genome structure by linking DNA sequence to transcriptional condensates and chromatin organization through specific transcription factors.

Compound-specific DNA adduct profiling with nanopore sequencing and IonStats

This study develops IonStats, a toolkit for profiling DNA adducts using nanopore sequencing, revealing compound-specific effects on sequencing data.

Blood donor biobank pipeline to collect genome-based samples for research

The paper describes a pipeline for collecting high-quality blood samples from donors for multi-omics research, demonstrating their suitability for genetic and phenotypic studies.

Whole Exome Sequencing and Single-Cell DNA Sequencing for Assessment of Clonal Heterogeneity and Evolution in Acute Myeloid Leukemia

The paper compares whole exome sequencing and single-cell DNA sequencing in AML to assess clonal heterogeneity and evolution, finding both methods have strengths and limitations in variant detection and clonal analysis.

Pathogenicity evaluation of coding germline variants identifies rare alleles enriched in hematological patients of a founder population

The paper evaluates germline variants in hematological malignancies using a computational pipeline, identifying pathogenic alleles and enriched variants in CHEK2 and other genes.

Single-cell transcriptomes identify patient-tailored therapies for selective co-inhibition of cancer clones

The paper develops a machine learning model to identify patient-specific multi-targeting therapies using single-cell transcriptomic data for hematological cancers and solid tumors.

Image-based profiling of healthy donor immune cells with Blood Cell Painting reveals novel genotype-phenotype associations

The paper develops Blood Cell Painting (BCP) to profile immune cell morphology in healthy donors, identifying genetic and clinical associations with immune cell features.

ArcheD, a residual neural network for prediction of cerebrospinal fluid amyloid-beta from amyloid PET images

ArcheD predicts cerebrospinal fluid amyloid-beta levels from amyloid PET images with strong correlation, showing distinct brain region contributions across AD stages.

A FinnGen pilot clinical recall study for Alzheimer's disease

The paper demonstrates the feasibility of conducting clinical recall studies from the FinnGen biobank to collect customized clinical data from participants with Alzheimer's disease or mild cognitive disorder.

Mutation-Attention (MuAt): deep representation learning of somatic mutations for tumour typing and subtyping

The paper introduces MuAt, a deep learning model that learns representations of somatic mutations to predict tumour types and subtypes with high accuracy.

DeepIFC: virtual fluorescent labeling of blood cells in imaging flow cytometry data with deep learning

DeepIFC uses deep learning to generate fluorescent marker images from imaging flow cytometry data, enabling accurate cell typing and recognition of unseen cell types.

Comparative analysis of molecular representations in prediction of drug combination effects

The paper compares rule-based and data-driven molecular representations for predicting drug combination effects using standardized screening data from 112 cancer cell lines.

Retrotransposon insertions can initiate colorectal cancer and are associated with poor survival

This study identifies retrotransposon insertions in colorectal tumors, linking them to cancer initiation and poor patient survival.

BasePlayer: Versatile Analysis Software For Large-Scale Genomic Variant Discovery

BasePlayer is a software tool for analyzing large-scale NGS data, enabling variant analysis, visualization, and integration of genomic data.

On enhancing variation detection through pan-genome indexing

The paper enhances variation detection by using pan-genome indexing to create a recombined reference sequence from multiple alignments, improving detection on real datasets.