Blood proteins have provided essential insights into how humans responded to the recent pandemic. To expand our understanding beyond patients seeking medical care, we conducted a citizen-centric survey with 2,000 random residents (age: 18–69 years) from Sweden's two largest cities in 2021. With self-sampled dried blood spots (DBS) and health information from 437 (22%) volunteers, we performed multi-analyte COVID-19 serology, measured autoantibodies (AAbs) against 22 interferons, and quantified 502 circulating low-abundant immune-related blood proteins. Antibody assays confirmed self-reported infections (26%) and vaccinations (40%), showed timing-dependent discrepancies in the immune response, and revealed anti-type I interferon AAbs co-occurring frequently alongside natural infections. Proteomics data added plausible mechanistic insights into cell-mediated processes: data-driven analyses revealed 24% of participants presented deviating immune phenotypes linked to infections, immunity, respiratory effects, and age. Multi-molecular DBS analysis of random layperson samples captured the broader spectrum of immune system states, adding relevant insights for clinical and public health investigations.

Motivation: Exploration of large-scale biological datasets remains a central challenge in computational biology. While many tools are available, they are often developed in isolation, leading to fragmented workflows, duplicated efforts, and limited reproducibility. There is a pressing need for flexible, standardized solutions that unify exploratory data analysis and biomarker discovery across diverse platforms.

Results: We present HDAnalyzeR, a user-friendly and extensible R package for the streamlined analysis of high-dimensional biological data. HDAnalyzeR provides modular, reproducible workflows that support a range of analyses, from quality control and dimensionality reduction to differential expression and enrichment analysis. The package features built-in visualization, metadata-aware modeling, and seamless integration with interactive apps and learning resources. We also present two case studies, where HDAnalyzeR dramatically reduced analysis time and code complexity while providing biologically meaningful insights, such as classification of blood cancer types with AUC = 1.0 and identification of thousands of solid tumor-associated genes. HDAnalyzeR is designed to support both beginner users and experienced bioinformaticians, promoting transparency, reproducibility, and publication-quality output.

Availability and implementation: HDAnalyzeR is freely available both as an open-source R package at https://github.com/kantonopoulos/HDAnalyzeR and a web application at https://hdanalyzer.serve.scilifelab.se.

Proteomic research enhances our understanding of health- and disease-related biological processes. Protein profiling during healthy childhood provides important insights into normal physiological development. We longitudinally measured 5416 plasma proteins at four follow-ups during childhood (4-, 8-, 16 years) and early adulthood (24 years) in 100 randomly selected subjects participating in a population-based Swedish cohort, using Olink Explore HT. In total, 3509 proteins were included in the analysis. 54% of the proteins were found to be associated with age, and we observed several protein trajectories from childhood to adulthood based on clustering. In addition to proteins involved in bone, teeth and cartilage formation, we identified differences in proteins involved in neural function, drug metabolism, and hormonal control. There were pronounced sex-related differences in protein levels, particularly at follow-ups 16 and 24, characterized by, for example, growth, response to stimuli and regulation of catabolic processes. We demonstrate dynamic age- and sex-related changes in protein levels during the first two decades of life. Our study results may serve as an important resource in understanding human physiological development, disease etiology, and for future protein biomarker research.

Achieving minimally invasive and rapid detection is a crucial goal in modern medicine. The comprehensive characterization of the blood proteome holds great promise in advancing our understanding of disease etiology, facilitating early diagnosis, risk stratification, and improved monitoring across various diseases and their subtypes. In this study, we collected plasma proteomes from over 3000 patients, representing 23 distinct diseases, encompassing a total of 1462 proteins. Based on histological knowledge, we developed a two-stage hierarchical multi-disease classifier and applied it to perform multi-disease classification on the collected proteomic data. Our results demonstrate that this empirically guided two-stage hierarchical multi-disease classifier outperforms traditional machine learning algorithms in terms of prediction performance, showing better balance and more meaningful feature selections. This finding highlights the positive role that domain expertise can play in machine learning-based disease detection, and underscores the potential of plasma proteomics for multi-disease screening.

Background & Aims: Primary sclerosing cholangitis (PSC) is a rare cholestatic liver disease with heterogeneous phenotypes and progression. Autoimmune traits, such as the presence of autoantibodies, are suspected to drive its heterogeneity. Methods: We performed a proteome-scale autoantibody screen of IgG and IgA isotypes using >42,100 protein fragments. This was followed by a validation of 1,153 selected autoantibodies, in serum samples from 466 patients with PSC in a longitudinal setting using the SUPRIM cohort and 214 controls. Results: We identified autoantibodies associated with clinical phenotypes, biochemical and clinical severity, comorbidities, and disease progression (e.g. alkaline phosphatase and albumin level p <e-10, presence of hepatobiliary malignancies p <0.001, seroconversion before transplantation p <0.001). Rather than a single universal autoantibody marker, small patient subgroups were positive for various autoantibodies with variable specificity. Global analysis of autoantigen targets revealed an overrepresentation of proteins normally expressed in immune-privileged sites, including the brain, testis, and retina. When interrogating tissue-specific autoantigen co-expression linked to expression and splicing quantitative trait loci of PSC risk variants, the thyroid emerged as an additional relevant tissue. We also detected increased autoantibody diversity associated with PSC duration and end-stage disease, already observable several years before liver transplantation. Multiomics analysis across body compartments confirmed neuroendocrine dysregulation in PSC. Our results are provided as a resource for further studies. Conclusions: Overall, our data support the cryptic antigen and epitope-drifting autoimmune theories and indicate that neuroendocrine dysregulation may contribute to PSC pathogenesis. Impact and implications: From a proteome-scale profiling of the SUPRIM cohort, we provide a short list of autoantibodies associated with clinical phenotypes and progression, along with the peptide sequences used to capture them. We identify across multiple datasets neuroendocrine deregulations in primary sclerosing cholangitis and provide a short list of related key plasma proteins. These data and technical details should facilitate validation studies, investigations of related pathophysiological mechanisms and development of low-cost tools for diagnostic or prognostic purposes.

Importance Identifying early proteomic profiles in infants who develop severe retinopathy of prematurity (ROP) may reveal targets for preventive interventions to reduce retinal vessel loss and the subsequent risk of severe ROP. Objective To assess early longitudinal profiles of blood protein levels in preterm infants with or without severe ROP and the effect of arachidonic acid (AA) and docosahexaenoic acid (DHA) supplementation. Design, Setting, and Participants This was an exploratory, post hoc analysis of serum proteome profiles in preterm infants in the double-masked Mega Donna Mega (MDM) randomized clinical trial using targeted Olink Proximity Extension Assay proteomics covering 538 analytes. The setting was 3 university hospitals in Sweden and included extremely preterm infants born before 28 weeks of gestational age (GA), from 2016 to 2019. Data were analyzed from January to March 2025. Exposures All infants received standard nutrition; additionally, half received enteral lipid supplementation with AA/DHA (100/50 mg/kg per day) from birth to term equivalent age. Main Outcomes and Measures Longitudinal protein profiles during the first month of life were examined using mixed models for repeated measures, adjusted for GA, study center, and AA/DHA supplementation, and tested for the interaction between severe ROP (stage ≥3 and/or treated) and postnatal age. Results A total of 177 extremely preterm infants (mean [SD] GA, 25.6 [1.4] weeks; 100 male [56.5%]) were included, of whom 50 (28.2%) developed severe ROP. Of 538 longitudinal analyzed proteins, 109 protein profiles in the first month of life associated with severe ROP, proteins related to immune response, apoptotic processes, blood coagulation, and lipid metabolism. The most pronounced association with severe ROP was a fast rise in fibroblast growth factor 21 (FGF-21; β = 0.68; 95% CI, 0.39-0.97; Q =.002) and tissue plasminogen activator (tPA; β = 0.21; 95% CI, 0.13-0.29; Q <.001) during the first postnatal days. The increase in serum FGF-21 level in the first week of life was associated with lower GA, lower birth weight, low enteral energy intake, and more days receiving mechanical ventilation. No association was observed between AA/DHA supplementation and the proteome. Conclusions and Relevance In this post hoc exploratory analysis of data from the MDM randomized clinical trial, a fast rise in FGF-21 levels, a metabolic stress-induced hormone, during the first postnatal days was strongly associated with the development of severe ROP in extremely preterm infants. These findings suggest that early interventions improving bioenergetic status may help prevent severe ROP.

The circulating proteome is a dynamic and accessible window into the biological state of the human body, reflecting its physiological and pathological processes. Advances in technologies to measure the plasma proteome now enable the measurement of thousands of proteins at population scale, opening new opportunities for the discovery of clinically relevant and minimally invasive biomarkers. These approaches hold promise for improving disease detection, patient stratification, and disease monitoring, positioning plasma proteomics at the forefront of precision medicine. Key to these advances are bioinformatic methods that identify candidate proteins associated with specific health and disease states from high-dimensional datasets. Despite the efforts combining large-scale proteomics with computational analyses, relatively few biomarkers have translated into clinical practice. This highlights the need for investigations that incorporate diverse cohorts and expand on classic comparisons against healthy controls, alongside an increased focus on validation strategies.

This thesis contributes to biomarker discovery by broadening the biological contexts that are profiled and compared. The first studies focus on cancer, starting by predicting the presence of cancer in patients with non-specific symptoms in Paper I, and identifying a protein panel able to distinguish between twelve cancer types in Paper II. Building on these findings, Paper III provides a deeper perspective of the circulating proteome across healthy individuals, during development, adulthood and aging, and a wide range of diseases. This is followed by Paper IV, which focuses on comparing the two main affinity proteomics platforms by assessing their complementarity and applicability in biomarker studies. Finally, the analysis of these large-scale datasets led to the development of streamlined bioinformatics pipelines, which are presented as an open-access package in Paper V.

Together, this work illustrates the potential of combining affinity proteomics with bioinformatics pipelines to profile the circulating proteome and derive biological insights. This thesis focuses on pan-disease comparisons, evaluates the complementarity of affinity proteomics platforms, and highlights the importance of reproducible biomarker discovery workflows. Developed within the framework of the Human Disease Blood Resource, the resulting data and insights are integrated into the Human Protein Atlas (www.proteinatlas.org), providing a resource for precision medicine research.

Proteiner i blodplasma återspeglar de fysiologiska och patologiska processer som sker i kroppen och utgör därmed en unik källa till information om människokroppens hälsotillstånd. Den tekniska utvecklingen inom analys av plasmaproteomet har möjliggjort att tusentals proteiner kan kvantifieras i tusentals prover. Detta har skapat goda förutsättningar för identifiering av nya kliniskt relevanta biomarkörer som kan mätas i ett enkelt blodprov och användas inom diagnostik, riskstratifiering och prognos. Tillämpningen av bioinformatiska metoder på högdimensionell data har varit en nyckel till identifiering av proteiner kopplade till specifika hälso- och sjukdomstillstånd. Trots omfattande metodutveckling inom både bioinformatik och olika proteomikplattformar har få biomarkörer introducerats inom klinisk kemi, vilket understryker behovet av bredare studier, utökande jämförelsegrupper och ett större fokus på validering.

Syftet med denna avhandling är att bidra till nya lovande biomarkörspaneler genom att skräddarsy och optimera de sammanhang där blodplasmaproteomet studeras. De inledande studierna fokuserar på cancerdiagnostik: i Artikel I identifieras biomarkörspaneler som kan ge ledtrådar om tidig utvecklad cancer hos patienter med diffusa symptom, och i Artikel II identifieras proteiner med potential att urskilja tolv olika cancerformer från varandra. Artikel III bygger vidare på de tidigare nämnda studierna genom att även inkludera friska vuxna som kombinerats med uppföljningsstudier hos barn som växer upp, äldre samt ett stort antal sjukdomstillstånd. Detta följs upp med Artikel IV som undersöker hur väl olika affinitetsproteomikplattformar stämmer överens, samt hur resultatet översätts till biologisk kontext. Slutligen presenterar Artikel V ett programmeringsbibliotek som innehåller flera analysmetoder för att effektivisera biomarkörsforskning baserad på storskalig data.

Sammantaget visar avhandlingen hur affinitetsproteomik och bioinformatiska arbetsflöden kan kombineras för att utforska det cirkulerande proteomet i olika hälso- och sjukdomstillstånd. Det övergripande arbetet belyser värdet av sjukdomsöverskridande jämförelser samt vikten av reproducerbara arbetsflöden vid biomarkörsstudier. Avhandlingen har genomförts inom ramen för Human Disease Blood Resource, som är en del av Human Protein Atlas (www.proteinatlas.org) där data och resultat som genererats som en resurs inom precisionsmedicin.

The human blood proteome provides a holistic readout of health states through the assessment of thousands of circulating proteins. In this study, we present a pan-disease resource to enable the study of diverse disease phenotypes within a harmonized proteomics dataset. By profiling protein concentrations across 59 diseases and healthy cohorts, we identified proteins associated with age, sex, and body mass index, as well as disease-specific signatures. This study highlights shared and distinct protein patterns across conditions, demonstrating the power of a unified proteomics approach to uncover biological insights. The dataset, covering 8262 individuals and up to 5416 proteins, serves as an online resource for exploring disease-specific protein profiles and advancing precision medicine research.

New technologies enable single-cell transcriptome analysis, mapping genome-wide expression across the human body. Here, we present an extended analysis of protein-coding genes in all major human tissues and organs, combining single-cell and bulk transcriptomics. To enhance transcriptome depth, 31 tissues were analyzed using a pooling method, identifying 557 unique cell clusters, manually annotated by marker gene expression. Genes were classified by body-wide expression and validated through antibody-based profiling. All results are available in the updated open-access Single Cell Type section of the Human Protein Atlas for genome-wide exploration of genes, proteins, and their spatial distribution in cells.