Statistical Analysis of Protein Microarray Data

Microarray data analysis represents one of the clearest examples of the highly beneficial interaction between bioinformatics and statistics. Protein microarrays are powerful tools for high-throughput studies of the human proteome; however, both systematic and non-systematic sources of bias can limit the optimal interpretation and ultimate utility of the data. In a typical protein microarray experiment, the number of samples is often limited, while the number of features in the raw data can exceed 60,000. In the case under consideration, there are 10 protein microarray datasets—one for each patient—comprising 5 patients diagnosed with Desmoplastic Small Round Cell Tumor (DSCRT) and 5 healthy controls, with each dataset containing 52,736 rows. To extract meaningful information from this high-dimensional data, various data pre-processing and statistical inference techniques will be employed. In particular, different methods for background correction and normalization will be evaluated to reduce technical noise and ensure robust data analysis. Subsequently, non-parametric tests will be applied for statistical inference, followed by p-value correction to account for multiple comparisons. The goal is to identify biomarkers that can highlight differences between diseased and healthy patients.