EviStat

Big data plays a crucial role in biomedical research, offering a wealth of information for various needs. However, for small to mid-size companies or research groups, collecting and profiling hundreds of clinical or experimental samples can be challenging. Our website provides a concise summary table of public (and sometimes very big) datasets, ready for immediate use. Each dataset has been downloaded, pre-formatted, and is accessible for mining knowledge relevant to your research goals. Consider re-analyzing these datasets to e.g. identify and cross-validate drug targets or biomarkers. We can further discuss and customize the analysis to meet your specific requirements. Explore the datasets here: https://lnkd.in/dK5_FKka #BigData #Omics #BiomedicalResearch #ResearchTools #DataAnalysis #PublicDatasets #Validation #DrugTarget #Biomarker

We often encounter a paper about new biomarkers for early diagnostics or disease prognosis, identified through omics analysis, e.g. gene expression data. However, most of these candidates never make it to clinical practice. Beyond obvious organizational and laboratory obstacles, there are hidden statistical reasons to that: 1)     The ever-growing abundance of molecular features (genes, proteins, SNPs, lncRNAs etc.) compared to the still limited number of patient samples, which may result in the selection of genes that merely coincide with the disease profile. 2)     The trivial multiple testing problem. Indeed, any long p-value series would contain seemingly significant “discoveries”; e.g. p<0.01 would occur in 1% of the cases. These would be actually false, since they emerge with comparable frequency even in random datasets, so that, say, 200 genes declared as differentially expressed. Publishing findings without adjusting p-values for multiple testing (as well as publication bias and other factors) leads to the situation when “Most Published Research Findings Are False”, as was brilliantly explained by John P. A. Ioannidis 19 years ago (https://lnkd.in/dMA6gYjn). Surprisingly many authors are still not aware of that(!) 3)     Lack of true validation. While most naïve publications are limited to discovery and validation within the same dataset, more advanced works employ public datasets for “external” validation. However, an uncontrolled "massaging" of the available datasets and missing important covariates could lead to eventually seeing “novel” markers. For a reviewer of submitted manuscripts, seeing the avalanche of such papers may feel desperate. Being rejected in one place, authors can easily turn to many others and enjoy the publication. Although, awareness of these challenges is crucial for real clinical innovation in biomarker research.   Read more about the basics of biomarker discovery and the essential requirements for building models in the provided resources: - Biomarker discovery concepts: https://lnkd.in/dtTiEh8U - Model building guidelines: https://lnkd.in/dPExr_Gj - A study of publication bias and p-hacking: https://lnkd.in/dkaKz9rf

Modern anti-cancer treatments are effective in only a minority of patients, with known determinants of treatment response offering only partial explanations. Attempts to enhance this knowledge using omics data have faced challenges. Our approach leverages network-level analysis and fuzzy set theory to develop a novel generation of biomarkers. The method, NeaMarker maps patient-specific gene signatures to characteristic pathways, creating diagnostic panels for predicting treatment outcomes. Learn more here: https://lnkd.in/dAKEe-dA #cancerresearch #biomarkers #precisionmedicine #TreatmentResponse #networks #NetworkAnalysis

Finding differential expressed genes or proteins can be challenging, especially when dealing with complex research questions like identifying molecular determinants of disease relapse years ahead. In the worst case no significant hits could be found at all. But fear not, there's a solution: utilize Network Enrichment Analysis (NEA) to uncover pathways associated with your phenotype of interest. Unlike traditional methods, NEA is highly sensitive and specific, detecting compact, functionally relevant signaling and regulatory events. NEA has proven invaluable in a number of studies, such as demonstrating drug activity, identifying drug resistance signatures, and pinpointing cytokines specific to autoimmune pathologies. Interested in learning more about NEA https://lnkd.in/gwqeRcWJ and its applications? Check out the success stories by the links below: - NEA in anti-cancer drug activity by Sprint Bioscience: https://lnkd.in/gMHvBwsi - NEA in drug resistance signatures by Karolinska Institutet: https://lnkd.in/gjtNAPmV - NEA in key cytokine discovery for autoimmune pathologies by Umeå University: https://lnkd.in/gsccptnk The possibilities with NEA and other methods are vast. Reach out to explore more efficient tools and resources for your research endeavors. #NetworkAnalysis #Biomarkers #PreclinicalStudies #NEA #PathwayAnalysis #Bioinformatics #Statistics #DifferentialExpression #Networks

EviStat is a consultancy offering biostatistics and bioinformatics services. By using omics and drug sensitivity datasets, EviStat can assist with drug target identification, discovery of drug response biomarkers, or experiment planning. The analysis can be expanded drastically by a rich public data collection, which was pre-processed and is stored locally. We possess multi-year, credit-rich experience in biomedical data analysis, distinguished by adequate, evidence-driven statistical approach. With multi-year experience and an evidence-driven statistical approach, we deliver quality and credit-rich solutions. #Biostatistics #Bioinformatics #DataAnalysis #Omics #Biomarkers

Happy to share the results of Sprint Bioscience latest study dissecting the effects of VPS34 inhibition in shaping the antitumor immune response. Through a close and stimulating collaboration with teams from academy (Bassam Janji (PhD, HDR), Katja Pokrovskaja Tamm) and industry (Madhumita B.) led by Yasmin Yu, PhD we report that our VPS34 inhibitor SB02024 induces the cGAS/STING pathway and greatly improves the efficacy of a STING agonist in a murine melanoma in vivo. Such deeper mechanistic understanding of the role of VPS34 in modulating type I interferon signaling paves the way to investigate combination therapies with different classes of immune-activating agents. Read more in Molecular Oncology:

It is time for me to advertise the second instance of the course “Scientific data analysis: from statistics to data science and back”. Starting in February 2024, it is part of the curriculum at The Free University (Brīvā universitāte) https://lnkd.in/dyxHVEC6 – a largely on-line school created by academics who had to leave Russian universities due to late developments, i.e. the war of Russia against Ukraine. The course would address basic concepts of statistical and data analyses and cover details of what I am best at: methods based on information, set, and graph theories. There are no tuition fees for students, and anyone can apply to any of the 50 courses by more than 100 professors. The deadline for applications is approaching (21st of January), so please spread the message among your Russian-speaking colleagues.