High throughput single cell long-read sequencing analyses of same-cell genotypes and phenotypes in human tumors.

Single cell Nanopore sequencing of full-length RNAs (scNanoRNAseq) is transforming single cell multi-omics analysis, however it is computationally challenged and relying on paralleled short-read sequencing to curate errors. We developed scNanoGPS to calculate same-cell genotypes-phenotypes from scNanoRNAseq data and eliminate dependance on short-reads guidance. To test its accuracy, robustness and applications, we analyzed 6 single nuclei transcriptomes composited of 4 frozen tumors and 2 cancer cell lines. Our results showed that scNanoGPS accurately deconvoluted raw long-reads into single-cells and single-molecules without short-reads guidance and calculated same-cell gene expressions, isoforms, and mutations and copy numbers simultaneously for thousands of cells. In a kidney tumor, we can identify cell-type-specific alternatively spliced genes enriched in important tumorigenic pathways, in addition to expression levels. Further, we detected transcriptome-wide mutations of each cell-type, enabling direct cell-lineage (genotype) and cell-fate (phenotype) alignment to investigate tumor progression. Together, scNanoGPS addresses major computational challenges and largely simplifies experimental workflow of scNanoRNAseq. Overall design: Single nucleus suspensions of all samples were loaded onto 10X Genomics Chromium instrucment for barcoding with scRNAseq protocol. A portion of barcoded cDNAs were fragmented to perform paralleled 3'scRNAseq on Novaseq 6000. Another portion of full-length cDNAs were sent for Oxford Nanopore sequencing on PromethION.