Enhancing reproducibility of gene expression analysis with known protein functional relationships: The concept of well-associated protein.

Identification of differentially expressed genes (DEGs) is well recognized to be variable across independent replications of genome-wide transcriptional studies. These are often employed to characterize disease state early in the process of discovery and prioritize novel targets aimed at addressing unmet medical need. Increasing reproducibility of biological findings from these studies could potentially positively impact the success rate of new clinical interventions. This work demonstrates that statistically sound combination of gene expression data with prior knowledge about biology in the form of large protein interaction networks can yield quantitatively more reproducible observations from studies characterizing human disease. The novel concept of Well-Associated Proteins (WAPs) introduced herein-gene products significantly associated on protein interaction networks with the differences in transcript levels between control and disease-does not require choosing a differential expression threshold and can be computed efficiently enough to enable false discovery rate estimation via permutation. Reproducibility of WAPs is shown to be on average superior to that of DEGs under easily-quantifiable conditions suggesting that they can yield a significantly more robust description of disease. Enhanced reproducibility of WAPs versus DEGs is first demonstrated with four independent data sets focused on systemic sclerosis. This finding is then validated over thousands of pairs of data sets obtained by random partitions of large studies in several other diseases. Conditions that individual data sets must satisfy to yield robust WAP scores are examined. Reproducible identification of WAPs can potentially benefit drug target selection and precision medicine studies.

An integrated approach using orthogonal analytical techniques to characterize heparan sulfate structure.

Heparan sulfate (HS), a glycosaminoglycan present on the surface of cells, has been postulated to have important roles in driving both normal and pathological physiologies. The chemical structure and sulfation pattern (domain structure) of HS is believed to determine its biological function, to vary across tissue types, and to be modified in the context of disease. Characterization of HS requires isolation and purification of cell surface HS as a complex mixture. This process may introduce additional chemical modification of the native residues. In this study, we describe an approach towards thorough characterization of bovine kidney heparan sulfate (BKHS) that utilizes a variety of orthogonal analytical techniques (e.g. NMR, IP-RPHPLC, LC-MS). These techniques are applied to characterize this mixture at various levels including composition, fragment level, and overall chain properties. The combination of these techniques in many instances provides orthogonal views into the fine structure of HS, and in other instances provides overlapping / confirmatory information from different perspectives. Specifically, this approach enables quantitative determination of natural and modified saccharide residues in the HS chains, and identifies unusual structures. Analysis of partially digested HS chains allows for a better understanding of the domain structures within this mixture, and yields specific insights into the non-reducing end and reducing end structures of the chains. This approach outlines a useful framework that can be applied to elucidate HS structure and thereby provides means to advance understanding of its biological role and potential involvement in disease progression. In addition, the techniques described here can be applied to characterization of heparin from different sources.

Multiomic study of skin, peripheral blood, and serum: is serum proteome a reflection of disease process at the end-organ level in systemic sclerosis?

BACKGROUND: Serum proteins can be readily assessed during routine clinical care. However, it is unclear to what extent serum proteins reflect the molecular dysregulations of peripheral blood cells (PBCs) or affected end-organs in systemic sclerosis (SSc). We conducted a multiomic comparative analysis of SSc serum profile, PBC, and skin gene expression in concurrently collected samples. METHODS: Global gene expression profiling was carried out in skin and PBC samples obtained from 49 SSc patients enrolled in the GENISOS observational cohort and 25 unaffected controls. Levels of 911 proteins were determined by Olink Proximity Extension Assay in concurrently collected serum samples. RESULTS: Both SSc PBC and skin transcriptomes showed a prominent type I interferon signature. The examination of SSc serum profile revealed an upregulation of proteins involved in pro-fibrotic homing and extravasation, as well as extracellular matrix components/modulators. Notably, several soluble receptor proteins such as EGFR, ERBB2, ERBB3, VEGFR2, TGFBR3, and PDGF-Rα were downregulated. Thirty-nine proteins correlated with severity of SSc skin disease. The differential expression of serum protein in SSc vs. control comparison significantly correlated with the differential expression of corresponding transcripts in skin but not in PBCs. Moreover, the differentially expressed serum proteins were significantly more connected to the Well-Associated-Proteins in the skin than PBC gene expression dataset. The assessment of the concordance of between-sample similarities revealed that the molecular profile of serum proteins and skin gene expression data were significantly concordant in patients with SSc but not in healthy controls. CONCLUSIONS: SSc serum protein profile shows an upregulation of profibrotic cytokines and a downregulation of soluble EGF and other key receptors. Our multilevel comparative analysis indicates that the serum protein profile in SSc correlates more closely with molecular dysregulations of skin than PBCs and might serve as a reflection of disease severity at the end-organ level.

Combining measurements to estimate properties and characterization extent of complex biochemical mixtures; applications to Heparan Sulfate.

Complex mixtures of molecular species, such as glycoproteins and glycosaminoglycans, have important biological and therapeutic functions. Characterization of these mixtures with analytical chemistry measurements is an important step when developing generic drugs such as biosimilars. Recent developments have focused on analytical methods and statistical approaches to test similarity between mixtures. The question of how much uncertainty on mixture composition is reduced by combining several measurements still remains mostly unexplored. Mathematical frameworks to combine measurements, estimate mixture properties, and quantify remaining uncertainty, i.e. a characterization extent, are introduced here. Constrained optimization and mathematical modeling are applied to a set of twenty-three experimental measurements on heparan sulfate, a mixture of linear chains of disaccharides having different levels of sulfation. While this mixture has potentially over two million molecular species, mathematical modeling and the small set of measurements establish the existence of nonhomogeneity of sulfate level along chains and the presence of abundant sulfate repeats. Constrained optimization yields not only estimations of sulfate repeats and sulfate level at each position in the chains but also bounds on these levels, thereby estimating the extent of characterization of the sulfation pattern which is achieved by the set of measurements.

Analyzing protein lists with large networks: edge-count probabilities in random graphs with given expected degrees.

We present an analytical framework to analyze lists of proteins with large undirected graphs representing their known functional relationships. We consider edge-count variables such as the number of interactions between a protein and a list, the size of a subgraph induced by a list, and the number of interactions bridging two lists. We derive approximate analytical expressions for the probability distributions of these variables in a model of a random graph with given expected degrees. Probabilities obtained with the analytical expressions are used to mine a protein interaction network for functional modules, characterize the connectedness of protein functional categories, and measure the strength of relations between modules.