Fig. 1

Feature generation for proposed models. a The transcript sequences were retrieved from RefSeq and were formed as a consensus matrix. b Each patient’s germline variants were embedded in the consensus matrix, forming a germline raw sequence for each sample. The brown dots are the germline variants including polymorphisms, deletions, and insertions. As an illustration, single nucleotide polymorphisms were identified and embedded in transcript A, E, and H. An in-frame shift deletion was embedded in transcript B and an in-frame shift insertion was embedded in transcript C. A frame shift deletion and frame shift insertion is embedded in transcripts D and E, respectively. Transcript F and G remained the same. c Each patient’s somatic mutations were embedded in the germline raw sequence (from B), forming a germline and cancer raw sequences. The green dots are the somatic mutations including SNVs, insertions, and deletions. As an illustration, the tissue gained somatic mutations in transcript A and E; gained a stop loss in transcript F; and gained a deletion that shifted the frame in transcript G