Distinguished Speaker Series: De novo transcriptome reconstruction from long reads

Abstract:  Long-read sequencing of transcripts with PacBio Iso-Seq and Oxford  Nanopore  Technologies  has  proven  to  be  central  to  the  study of  complex  isoform  landscapes  in  many  organisms.  However,  current de  novo transcript  reconstruction  algorithms  from  long-read  data  are limited,  leaving  the  potential  of  these  technologies  unfulfilled.  A  common  bottleneck  is  the  dearth  of  scalable  and  accurate  algorithms  for clustering  long  reads  according  to  their  gene  family  of  origin.  A second bottleneck is to be able to distinguish sequences errors from true variation with these families. I will present two recent methods to address these challenges. The first is IsoCon (Sahlin et al, 2018, Nat Comm),  a method to determine the full-length transcripts of multicopy gene families at nucleotide-level precision, from PacBio data. I will show how IsoCon was applied to Y chromosome ampliconic gene families, each of which contains many nearly identical gene copies. The second is isONclust (Sahlin & Medvedev, RECOMB 2019), a clustering algorithm that can assign Nanopore reads to their gene family of origin. ​

Host: Prof. Ron Shamir, Computer Science School, TAU.