Integration of single cell multi-omics data

Recent advances in data extraction technologies enable researchers to collect new types of data. Analyzing these data can provide a new level of understanding of cells and the mechanisms that underlies them. Multiple methods exist to analyze these data. Including factorization methods. They are intended to reduce the dimension and enlighten patterns in the data matrices. The integrative nonnegative matrix factorization with shared and unshared features (UINMF) method is a new matrix integration method that reduces the dimension and enlightens clusters in the data. It is derived from the workhouse method, in the field of matrix factorization, nonnegative matrix factorization (NMF). It is extended to capture different signals distinctly in cells and account for all features of the datasets, the features shared across datasets as well as the features unshared. The purpose of this work is to evaluate if the algorithm is reliable and can be used on real biological data. We will therefore test if the algorithm can recognize patterns and clusters among test datasets. The test datasets will be constructed such that they will have proprieties similar to real biological data. For instance a high number of samples (cells) and features (genes).