Biological protocols are terse, telegraphic recipes that document common wetlab tasks, for instance, sorting DNA molecules by length via gel electrophoresis, growing a liquid culture of E coli, or extracting plasmid DNA from E coli.

Protocols are generally written and shared in plain text with no machine-friendly structure, which is a shame, because all protocols have some elements in common: they specify functionally important inputs, outputs, and a series of biological operations that will transform the inputs into the outputs. Protocols are wetlab functions composed of a series of biological operations with well-defined inputs and outputs.

I've been discussing this in the context of wetlab workflow planning at the diybio google group. Here are some of my thoughts:

A simple way to play with structured protocols would be to establish a small controlled vocabulary for describing the inputs and outputs to various protocols. Meta-protocols could be created by searching for a path from a desired input to a desired output. For example, here is an input and output that would require stitching together several protocols:

Input: 10 ug 1uM plasmid DNA (BBa_I20259) Output: 400 Relative Fluorescence Units, 500nm.

The input / output controlled vocabulary is potentially very complex. I'm not sure how to deal with units.

Looking forward, if the vocabulary of available biological operations (verbs) and inputs/outputs (nouns) was structured in a machine-readable way, it could form the basis of a biological "test-driven development" system. First, desired biological outcomes would be expressed using the vocabulary. Then the biologist would execute an automatically-generated protocol until the tests passed.