Clinical and Microbiome Data Analysis
Data analysis and interpretation is a pillar of biomedical research even though respective expertise can be a bottleneck for basic and clinical research groups.
Our expertise spans from sample size calculations to multi-omics analysis. We find the best solution for each data set and adapt to new challenges. Together with our data science team we can also customize analysis pipelines tailor made for your data structure.
Clinical and Microbiome Data Analysis
Data analysis and interpretation is a pillar of biomedical research even though respective expertise can be a bottleneck for basic and clinical research groups.
Our expertise spans from sample size calculations to multi-omics analysis. We find the best solution for each data set and adapt to new challenges. Together with our data science team we can also customize analysis pipelines tailor made for your data structure.
Sample size calculations
Estimating the correct sample size for a planned study can be key to balance necessary effort and costs but is also an ethical and data protection issue. Straight forward sample size calculations can be easily done on web-based platforms without specific background knowledge. At CBmed, we use in-silico simulations to tackle also the more complicated cases, where endpoints assessments require more complex statistical models. In cooperation with our clinical experts, we find the most probable effect size upon which we can estimate the most sensible sample size.
Data Quality
Collecting data in the highest quality possible is the best foundation for sound data analysis. At CBmed, we established the use of electronic case report form to ensure quality and reduce the error rate during data collection. The structured data collection forms check the data already while entering, provide valid data types for each item and therefore enable multi-operator data entry without quality loss.
Time-to-event analysis
In medicine, time dependent endpoints are commonly occurring and build the foundation for prognosis, prediction and risk assessment. Mortality, the occurrence or recurrence of a complication or the risk to contract a disease are typical endpoints whose analysis can be enriched by including a time component. Visit our publication hub to get a picture of our expertise in this field.
Multivariate regression models
Diseases are often multi-factorial in their pathophysiology and these factors are often interlinked. This might cause confounding in the statistical analysis and might result in misleading results. At CBmed, we use multivariate regression models for identifying independent predictors and estimate their independent effect on the outcome. When building regression models, the right selection of predictors is key to a slim and efficient model. We employ a variety of selection algorithms such as backward elimination, forward or stepwise selection in building the most effective and relevant model. In some cases, it is necessary to use regularized regression, specifically if when we deal with high throughput data where the variables likely outnumber the observations. Our experts are skilled in finding the best model for each data set and applying tailor-made data analysis solutions to each problem. See potential applications for these models in our latest publications.
Microbiome analysis
Microbiome analysis is central to our research activities. With our strong focus on the microbiome, in clinical settings but also with our ex-vivo microbiome model, we are highly skilled in the analysis of sequencing data, either from 16S rRNA gene sequencing or shot gun metagenomics. In addition to the well-known diversity assessments and compositional comparisons, we specialize in the analysis of longitudinal sequencing data. Prime applications for our workflows are clinical intervention studies but also multi-armed experiments with repeated measurements in our in-vitro microbiome model. A prime example would be our work on probiotic supplementation during mild COVID infection in home quarantine, where we monitored the alterations in the gut microbiome over a period of 30 days.
Although we focus on the intestinal microbiome and primarily work with stool samples, we also have experience with other sample material such as saliva, urine or organ tissue from human or murine origin. Please get in contact to learn more.
Metabolomics
The human and bacterial metabolome is a rich source of predictive and diagnostic biomarkers. At CBmed we tap into this target rich environment by analysing the metabolome in stool, serum or urine. Together with our partners, we can offer a variety of methods for targeted and untargeted metabolomics. The analysis of these high throughput data sets is supported by multivariate models, dimension reduction and machine learning algorithms. For more details, get in contact with our experts.
Multi-omics data analysis
The integration of different data sets can enhance the change to uncover underlying mechanisms. A clinical study can produce several large data sets containing information about the microbiome, metabolome from different compartments, clinical data, routine blood work, information about soluble biomarkers among others. Conventionally, these data sets are analyzed separately, and the results are interpreted together. With our data-driven multi-omics approach, we combine various data sets before analysis and inform our algorithms with parameters of all data pools to uncover predictive or discriminating biomarker combinations. Multi-omics analysis can also be used to assess the associations between two data sets, for example to bring the microbiome composition in relation to quality-of-life parameters.
