The Growing Need for Domain-Driven Languages
With computer science permeating every research disciple such that even rudimentary computational skills are now required for most research careers, it should not be the case that one needs advanced knowledge of computer science to be effective in their field.
If we take Bioinformatics as an example, the non-computational biologist practitioner should be able to express theories in a truncated form of how they already speak of biology -- a Domain-Driven Language (DDL) which can then be computationally processed. Tools such as Nextflow and Snakemake, as workflow managers, make progress in this direction, but as they do not abstract away the data flow model they rely upon, do not go far enough.
I posit that unified domain-driven grammars are possible consisting of: verbs which define the action to be done (e.g., align); adjectives which define how the verb is to be done (e.g., local); nouns which define what the verb acts upon (e.g., sequence 1 and sequence 2); intents which define manner of action (e.g., longest match); and helpers which are meant to clarify intent (e.g., sequence 1 is reference, sequence 2 is query). An executed DDL should be able to use the same grammar to describe what was run in terms the executor understands.
Published: 5 Sep, 2019