Disagree with converting Sequence to a definitional-only resource. Rationale: [1] We need somewhere to include the patient's observed sequence. This can be the Sequence resource, or need to add another attribute to Haplotype observation; [2] ONC's GACS scenario: Sequencing labs are finding the need to make much more sequencing data available. It's unclear what format this will take - maybe FASTQ, BAM, and VCF files. To expose this raw data via a FHIR interface, we'll need a just-in-time FHIR conversion algorithm, and it seems as though the Sequence resource is the best target because it can represent the observed sequence, the window tested, and phased variants within the window; [3] We need somewhere to indicate analytic accuracy of (variant and wildtype) base calls. This can be the Sequence resource, or we need to define another observation profile; [4] it is premature to talk about deprecating the Sequence resource until we've figured out where and how the contained functionality will reside.