Comparison of Fusion Methodologies Using CNV and RNA-Seq for Cancer Classification: A Case Study on Non-Small-Cell Lung Cancer

Lung cancer is one of the most frequent cancer types, and one among those causing more deceases worldwide. Nowadays, in order to improve the diagnosis of cancer more screenings are performed to the same patient and various biological sources are being gathered. Fusing the information provided by these sources can lead to a more robust diagnosis, which can improve the prognosis of the patient. In this work, a comparison of fusion methodologies (early and intermediate) using RNA-Seq and Copy Number Variation data for Non-Small-Cell Lung Cancer classification is performed. We found that great results can be attained using both fusion methodologies, with an AUC of 0.984 for the early fusion and 0.989 for the intermediate fusion, improving those obtained by each source of information independently (0.978 RNA-Seq and 0.910 Copy Number Variation). This work shows that fusion methodologies can enhance the classification of non-small-cell lung cancer, and that these methodologies can be promising for the diagnosis of other cancer types.