National Yang-Ming University + 886 2 2826 7391

Bring Data Science to Biological Research

Read more

Connecting data to scientific discovery

Read more

You can be sure of one thing.

Read more

Identifying drug targets for cancer stem cells using feature selection and gene expression data

Cancer stem cells (CSCs) play an important role in cancer development due to their abilities to initiate tumor and resist treatments, and therefore identifying gene expression differences between CSCs and non-CSCs may facilitate cancer therapies. In addition, metabolic reprograming has been proposed to contribute to tumor growth and cancer survival. This research aims to identify biomarkers in metabolic and other several pathways as the potential targets for cancer treatments.

Gene expression data (196 CSC and 196 non-CSC samples) were collected from Gene Expression Omnibus. To avoid feature selection bias, five distinct FS methods were applied to the data set. Furthermore, Connectivity Map (CMAP) was used to identify potential drugs targeting cancer stem cells.

Among the 46 genes that were selected by three or more FS methods, three of them are known to be involved in metabolic pathways: sphingolipid, pyrimidine and selenocompound metabolism. In particular, CGS is a known treatment target that has the ability in modulating drug resistance. Although two inhibitors (Miglustat and PDMP) have been proposed for CGS, no further application has been made due to low specificity and undesirable side effects. In addition, of the seven candidate drugs obtained from the CMAP analysis, the top one, the Camptothecin, exhibits the ability to overcome drug resistance.

By analyzing the expression data between CSC and non-CSC, we have demonstrated that the identified genes and their inhibitors may be considered as the potential biomarkers and treatments of drug resistance, respectively. Combined with traditional chemotherapy, the proposed drugs might help kill drug-resistant CSCs, thereby increase the survival rates of cancer.