Olink Statistical Services
Statistical analysis and data visualization
As part of Olink Proteomics’ commitment to provide our customers with the highest possible standard of products and services for targeted human protein biomarker discovery, we are delighted to offer our fee-for-service Statistical Services. Our team of biostatisticians can help you with customized statistical analysis to help ensure that you maximize the value and information output from your studies run using Olink panels. The goal for us is to help you with design and analysis decisions leading to powerful and efficient studies and useful results.
Overview of our biostatistical offering:
- Performed by Olink expert professionals experienced in handling this type of data
- Customizable according to client needs
- Provides a fast and reliable way to get the most out of your experiment
- Charged by an hourly rate, according to the scope of the project.
- Initial planning and study design as well as discussion of the results is included in the service
We perform different kinds of statistical analysis depending on the need of the customer and the available data. For example:
- Group comparisons using T-test, ANOVA or linear regression
- Analysis of timeseries or repeated measurements by linear mixed effect models
- Unsupervised data overview and cluster analysis (PCA and hierarchical cluster for example)
- Machine learning for feature selection and prediction/classification
- Annotation and enrichment
Some examples of the wide range of biostatistical support available are shown below.
Group comparisons presented in a volcano plot
This provides fast and easy testing of a single hypothesis for all proteins analyzed in the study. The test can establish associations between protein levels and the variable of interest. Available tests include, but are not limited to T-tests, ANOVA and linear regressions.
Results from a T-test can be presented in a volcano plot (you can read more about volcano plots here).
Example of a volcano plot with significance on the y-axis and difference on the x-axis. The color indicates significance (multiple testing corrected p-value<0.05) and the line indicates raw p-value = 0.05. Top 10 proteins are labelled with protein names.
Linear mixed effect models presented in a point-range plot
To study within patient protein level changes over time that is different between groups, linear mixed effect model can be used. Significant proteins are visualized in point-range plots.
Example of point-range plots with NPX on y-axis and Time points on x-axis. The points are the marginal mean estimated by the model and the vertical bars are the 95% confidence intervals. The color indicates the groups.
Data overview and patterns by hierarchical clustering
Hierarchical clustering can be used to get an overview of data, and to identify subgroups of similar samples or proteins based on protein profiles.
Example of a heatmap based on hierarchical clustering analysis where similar proteins and samples are placed next to each other and the protein level is represented by a color gradient.
Want to know more?
Contact your local sales representative for further information about our services, or if you have questions about data analysis please contact the Data Science team.