The　long　reads　of　Nanopore　sequencing　permit　accurate　transcript　assembly　and　ease　in　discovering　novel　transcripts　with　potentially　important　functions　in　cancers.　The　wide　adoption　of　Nanopore　sequencing　for　transcript　quantification,　however,　is　largely　limited　by　high　costs.　To　address　this　issue,　we　developed　a　bioinformatics　software,　NovelQuant,　that　can　specifically　quantify　long-read-assembled　novel　transcripts　with　short-read　sequencing　data.　Nanopore　Direct　RNA　Sequencing　was　carried　out　on　three　hepatocellular　carcinoma　(HCC)　patients＇　tumor,　matched　portal　vein　tumor　thrombus,　and　peritumor　to　reconstruct　the　HCC　transcriptome.　Then,　based　on　the　reconstructed　transcriptome,　NovelQuant　was　applied　on　Illumina　RNA　sequencing　data　of　59　HCC　patients＇　tumor　and　paired　peritumor　to　quantify　novel　transcripts.　Our　further　analysis　revealed　361　novel　transcripts　dysregulated　in　HCC　and　that　101　of　them　were　significantly　associated　with　prognosis.　There　were　19　novel　prognostic　transcripts　predicted　to　be　long　noncoding　RNAs　(lncRNAs),　and　some　of　them　had　regulatory　targets　that　were　reported　to　be　associated　with　HCC.　Additionally,　42　novel　prognostic　transcripts　were　predicted　to　be　protein-coding　mRNAs,　and　many　of　them　could　be　involved　in　xenobiotic　metabolism.　Moreover,　the　tumor-suppressive　roles　of　two　representative　novel　prognostic　transcripts,　CDO1-novel　(lncRNA)　and　CYP2A6-novel　(protein-coding　mRNA),　were　further　functionally　validated　during　HCC　progression.　Overall,　the　current　study　shows　a　possibility　of　combining　long-　and　short-read　sequencing　to　explore　functionally　important　novel　transcripts　in　HCC　with　accuracy　and　cost-efficiency,　which　expands　the　pool　of　molecular　biomarkers　that　could　enhance　our　understanding　of　the　molecular　mechanisms　of　HCC.