The most common methods to discover cancer-associated transcripts from sequencing data are feature-based. In these approaches, the sequenced RNAs are mapped to a reference genome and overlapped with gene annotations (features). This approach works well when the aim is to quantify previously annotated molecules. However, we anticipated that cancer-specific RNAs may arise from the aberrant transcription of normally silenced regions in cancer and therefore do not overlap with any known or annotated genomic features. In addition, transcripts unique to cancer may originate from genomic sequence aberrations that do not exist in the reference human genome, and thus cannot be mapped. Therefore, the traditional feature-based approaches might miss cancer-specific RNAs that play important roles in cancers or could function as useful biomarkers. In this project we address these challenges by applying and developing genome-free and sequence-centric approaches that treats each unique sequence as an independent entity, regardless of its genomic mapping or annotation status. This should allow us to discover cancer-specific novel transcripts in an unbiased way, allowing us to find new cancer biomarkers. This project is the main PhD project of my student Panagiotis Kalogeropoulos. The project is funded by Cancerfonden with project number 24 3710 Pj. We have already obtained permission to access the human sequence data from the dbGaP database, however since the data are sensitive, it is essential that we work with them in the secured Bianca/Castor environment.