2.1. YAML Guide#

YAML is a file format designed to be readable by both computers and humans. This guide introduces the features of YAML relevant when writing CWL descriptions and input parameter files.

Note

You can skip this section if you are already comfortable with YAML.

2.1.1. Contents#

2.1.2. Key-Value Pairs#

Fundamentally, a file written in YAML consists of a set of key-value pairs. Each pair is written as key: value, where whitespace after the : is optional. Key names in CWL files should not contain whitespace - We use camelCase for multi-word key names that have special meaning in the CWL specification and underscored key names otherwise. For example:

first_name: Bilbo
last_name:  Baggins
age_years:  111
home:       Bag End, Hobbiton

The YAML above defines four keys - first_name, last_name, age_years, and home - with their four respective values. Values can be character strings, numeric (integer, floating point, or scientfic representation), Boolean (true or false), or more complex nested types (see below).

Values may be wrapped in quotation marks but be aware that this may change the way that they are interpreted i.e. "1234" will be treated as a character string , while 1234 will be treated as an integer. This distinction can be important, for example when describing parameters to a command: in CWL all parts of baseCommand must be strings so, if you want to specify a fixed numeric value to a command, make sure that you wrap that numeric value in quotes: baseCommand: [echo, "42"].

2.1.3. Comments#

You may use # to add comments to your CWL and parameter files. Any characters to the right of # will be ignored by the program interpreting the YAML. For example:

first_name: Bilbo
last_name:  Baggins
age_years:  111
# this line will be ignored by the interpreter
home:       Bag End, Hobbiton # this is ignored too

If there is anything on the line before the comment, be sure to add at least one space before the #!

2.1.4. Maps#

When describing a tool or workflow with CWL, it is usually necessary to construct more complex, nested representations. Called maps, these hierarchical structures are described in YAML by providing additional key-value pairs as the value of any key. These pairs (sometimes referred to as “children”) are written on new lines under the key to which they belong (the “parent”), and should be indented with two spaces (⇥tab characters are not allowed). For example:

cwlVersion: v1.0
class: CommandLineTool
baseCommand: echo
inputs: # this key has an object value
  example_flag: # so does this one
    type: boolean
    inputBinding: # and this one too
      position: 1
      prefix: -f

The YAML above illustrates how you can build up complex nested object descriptions relatively quickly. The inputs map contains a single key, example_flag, which itself contains two keys, type and inputBinding, while one of these children, inputBinding, contains a further two key-value pairs (position and prefix). See the Arrays section below for more information about providing multiple values/key-value pairs for a single key. For comparison with the example YAML above, here is a graphical representation of the inputs object it describes.

graph TD inputs --> example_flag example_flag --> type type --- bool((boolean)) example_flag --> inputBinding inputBinding --> position inputBinding --> prefix position --- posval((1)) prefix --- prefval(('-f'))

2.1.5. Arrays#

In certain circumstances it is necessary to provide multiple values or objects for a single key. As we’ve already seen in the Maps section above, more than one key-value pair can be mapped to a single key. However, it is also possible to define multiple values for a key without having to provide a unique key for each value. We can achieve this with an array, where each value is defined on its own line and preceded by -. For example:

touchfiles:
  - foo.txt
  - bar.dat
  - baz.txt

and a more complex example combining maps and arrays:

exclusive_parameters:
  type:
    - type: record
      name: itemC
      fields:
        itemC:
          type: string
          inputBinding:
            prefix: -C
    - type: record
      name: itemD
      fields:
        itemD:
          type: string
          inputBinding:
            prefix: -D

2.1.6. JSON Style#

YAML is based on JavaScript Object Notation (JSON) and maps and arrays can also be defined in YAML using the native JSON syntax. For example:

touchfiles: [foo.txt, bar.dat, baz.txt] # equivalent to first Arrays example

and:

# equivalent to the `inputs` example in "Maps" above
inputs: {example_flag: {type: boolean, inputBinding: {position: 1, prefix: -f}}}

Native JSON can be useful to indicate where a field is being left intentionally empty (such as [] for an empty array), and where it makes more sense for the values to be located on the same line (such as when providing option flags and their values in a shell command). However, as the second example above shows, it can severely affect the readability of a YAML file and should be used sparingly.

2.1.7. Reference#

The Learn YAML in Y Minutes reference was very helpful for us while we wrote this guide, though it also covers features that are not valid in CWL.