I'm interested in know how one goes about creating a high-quality reference genome to align NGS data to, but I find it difficult to find detailed information about this.
The obvious process is (1) sequence an individual, preferably with high coverage, and (2) de novo assemble this sequence. In addition, some reference contain a "decoy sequence" with DNA from the Epstein-Barr virus. Multiple individuals are used in the GRCh38 reference, such that it represents a haploid mosaic (I'm unclear on what this means exactly).
The above is the extent of my knowledge about the creation of a reference genome. I would like to know more. Information on any reference genome build, not just GRCh38, would be appreciated.
Keep it mind that the original reference human genome was completed in 2003. Many people have been working on improving it ever since, so it's a lot of work.