Predicting Chordoma Metastasis Using Gene Expression Data and Machine Learning

Abstract

Abstract

Chordoma is a type of cancer that occurs in the skull and spine region. Every year 300 patients are diagnosed with Chordoma, while in 30-40% patients metastasis occurs. The Research is focused on the prediction of Chordoma Cancer Metas- tasis, using a gene expression data set generated through RNA Sequencing. The dataset is acquired from the OpenPBTA Platform. Different pre-processing steps are applied to the dataset, along with Feature Reduction techniques like PCA, which are employed to select the top features to be used for the training stage. After applying various machine learning algorithms SVM Classifier for Chordoma Metastasis performed better than its counterparts.

Keywords: Bio-informatics, Machine Learning, Cancer, Gene Expression