Fig. 2

Read mapping versus de novo genome assembly for variant discovery. A traditional approach uses long-read mapping to a reference genome to identify SNVs, indels, and SV signatures, while de novo genome assembly reconstructs the two haplotypes of the sequenced individual and permits the direct comparison of assemblies (in clinical settings, ideally, parents versus proband). Genome assembly improves variant discovery, as all types of variations are fully sequence resolved and do not have to be inferred from SV signatures. Moreover, using a reference genome such as GRCh38 introduces biases due to the incompleteness of certain regions and misassembled complex loci. De novo genome assembly is the approach that we expect to substitute all the others and eventually be the gold standard method for variant discovery. (Visualization of assembly comparison adapted from SafFire [90].)