Application of K-Means Clustering to Identify Similar Gene Expression Patterns during Erythroid Development
Heba Saadeh, Reem Q. Al Fayez, Basima Elshqeirat · International Journal of Machine Learning and Computing · 2020
Erythropoiesis is the specific lineage in which the haematopoietic stem cells (HSC) differentiate into red blood cells. During their development, HSC undergo global gene expression changes to reflect the current developmental stage needs.A good way to identify the set of genes that have similar global expression patterns across the different developmental stages is through clustering.Unsupervised clustering aims at highlighting these co-regulated genes without prior knowledge regards their full interactions.In this study, we apply k-means clustering on a gene expression microarray data that measures the expression levels of human genes at four erythropoiesis stages.Eight clusters have been identified; one cluster, in particular, of 450 genes (C4) is more active toward the maturation stages and it is involved in cell division and DNA replication processes, which are vital during development.Another cluster of 234 genes (C7) is involved in autophagy (cells consumption/destruction), which is known to be involved in enucleation (expulsion of the nucleus from the cell).