avworkflow: Workflow submissions and file outputs
In Bioconductor/AnVIL: Bioconductor on the AnVIL compute environment

Description Usage Arguments Details Value Examples

avworkflows() returns a tibble summarizing available workflows.

avworkflow_jobs() returns a tibble summarizing submitted workflow jobs for a namespace and name.

avworkflow_files() returns a tibble containing information and file paths to workflow outputs.

avworkflow_localize() creates or synchronizes a local copy of files with files stored in the workspace bucket and produced by the workflow.

avworkflow_configuration_template() returns a template for defining workflow configurations. This template can be used as a starting point for providing a custom configuration.

avworkflow_configuration() returns a list structure describing an existing workflow configuration.

avworkflow_import_configuration() updates an existing configuration, e.g., changing inputs to the workflow.

avworkflows(namespace = avworkspace_namespace(), name = avworkspace_name())

avworkflow_jobs(namespace = avworkspace_namespace(), name = avworkspace_name())

avworkflow_files(submissionId = NULL, bucket = avbucket())

avworkflow_localize(
  submissionId = NULL,
  destination = NULL,
  type = c("control", "output", "all"),
  bucket = avbucket(),
  dry = TRUE
)

avworkflow_configuration_template()

avworkflow_configuration(
  configuration_namespace,
  configuration_name,
  namespace = avworkspace_namespace(),
  name = avworkspace_name()
)

avworkflow_import_configuration(
  config,
  namespace = avworkspace_namespace(),
  name = avworkspace_name()
)

`namespace`	character(1) AnVIL workspace namespace as returned by, e.g., `avworkspace_namespace()`
`name`	character(1) AnVIL workspace name as returned by, eg., `avworkspace_name()`.
`submissionId`	a character() of workflow submission ids, or a tibble with column `submissionId`, or NULL / missing. See 'Details'.
`bucket`	character(1) name of the google bucket in which the workflow products are available, as `gs://...`. Usually the bucket of the active workspace, returned by `avbucket()`.
`destination`	character(1) file path to the location where files will be synchronized. For directories in the current working directory, be sure to prepend with `"./"`. When `NULL`, the `submissionId` is used as the destination. `destination` may also be a google bucket, in which case th workflow files are synchronized from the workspace to a second bucket.
`type`	character(1) copy `"control"` (default), `"output"`, or `"all"` files produced by a workflow.
`dry`	logical(1) when `TRUE` (default), report the consequences but do not perform the action requested. When `FALSE`, perform the action.
`configuration_namespace`	character(1) namespace of the workflow. Often the same as the namespace of the workspace. Discover configuration namespace and name information from `avworkflows()`.
`configuration_name`	character(1) name of the workflow, from `avworkflows()`
`config`	a named list describing the full configuration, e.g., created from editing the return value of `avworkflow_configuration()` or `avworkflow_configuration_template()`.

For avworkflow_files(), the submissionId is the identifier associated with the workflow job, and is present in the return value of avworkflow_jobs(); the example illustrates how the first row of avworkflow_jobs() (i.e., the most recenltly completed workflow) can be used as input to avworkflow_files(). When submissionId is not provided, the return value is for the most recently submitted workflow of the namespace and name of avworkspace().

avworkflow_localize(). type = "control" files summarize workflow progress; they can be numerous but are frequently small and quickly syncronized. type = "output" files are the output products of the workflow stored in the workspace bucket. Depending on the workflow, outputs may be large, e.g., aligned reads in bam files. See gsutil_cp() to copy individual files from the bucket to the local drive.

1
2
3

`avworkflow_localize()` treats `submissionId=` in the same way
as `avworkflow_files()`: when missing, files from the most
recent workflow job are candidates for localization.

avworkflows() returns a tibble. Each workflow is in a 'namespace' and has a 'name', as illustrated in the example. Columns are

name: workflow name.
namespace: workflow namespace (often the same as the workspace namespace).
rootEntityType: name of the avtable() used to retrieve inputs.
methodRepoMethod.methodUri: source of the method, e.g., a dockstore URI.
methodRepoMethod.sourceRepo: source repository, e.g., dockstore.
methodRepoMethod.methodPath: path to method, e.g., a dockerstore method might reference a github repository.
methodRepoMethod.methodVersion: the version of the method, e.g., 'main' branch of a github repository.

avworkflow_jobs() returns a tibble, sorted by submissionDate, with columns

submissionId character() job identifier from the workflow runner.
submitter character() AnVIL user id of individual submitting the job.
submissionDate POSIXct() date (in local time zone) of job submission.
status character() job status, with values 'Accepted' 'Evaluating' 'Submitting' 'Submitted' 'Aborting' 'Aborted' 'Done'
succeeded integer() number of workflows succeeding.
failed integer() number of workflows failing.

avworkflow_files() returns a tibble with columns

file: character() 'base name' of the file in the bucket.
workflow: character() name of the workflow the file is associated with.
task: character() name of the task in the workflow that generated the file.
path: charcter() full path to the file in the google bucket.

avworkflow_localize() prints a message indicating the number of files that are (if dry = FALSE) or would be localized. If no files require localization (i.e., local files are not older than the bucket files), then no files are localized. avworkflow_localize() returns a tibble of file name and bucket path of files to be synchronized.

avworkflow_configuration_template() returns a list providing a template for configuration lists, with the following structure:

namespace character(1) configuration namespace.
name character(1) configuration name.
rootEntityType character(1) or missing. the name of the table (from avtables()) containing the entitites referenced in inputs, etc., by the keyword 'this.'
prerequisites named list (possibly empty) of prerequisites.
inputs named list (possibly empty) of inputs. Form of input depends on method, and might include, e.g., a reference to a field in a table referenced by avtables() or a character string defining an input constant.
outputs named list (possibly empty) of outputs.
methodConfigVersion integer(1) identifier for the method configuration.
methodRepoMethod named list describing the method, with character(1) elements described in the return value for avworkflows().
- methodUri
- sourceRepo
- methodPath
- methodVersion. The REST specification indicates that this has type integer, but the documentation indicates either integer or string.
deleted logical(1) of uncertain purpose.

The exact format of the configuration is important.

One common problem is that a scalar character vector `"bar"` is
interpretted as a json 'array' `["bar"]` rather than a json
string `"bar"`. Enclose the string with
`jsonlite::unbox("bar")` in the configuration list if the
length 1 character vector in R is to be interpretted as a json
string.

A second problem is that an unquoted unboxed character string
`unbox("foo")` is required by AnVIL to be quoted. This is
reported as a warning() about invalid inputs or outputs, and
the solution is to provide a quoted string `unbox('"foo"')`.

avworkflow_configuration() returns a list structure describing the configuration. See avworkflow_configuration_template() for the structure of a typical workflow.

avworkflow_import_configuration() returns an object describing the updated configuration. The return value includes invalid or unused elements of the config input. Invalid or unused elements of config are also reported as a warning.

if (gcloud_exists() && nzchar(avworkspace_name()))
    ## from within AnVIL
    avworkflows() %>% select(namespace, name)

if (gcloud_exists() && nzchar(avworkspace_name()))
    ## from within AnVIL
    avworkflow_jobs()

if (gcloud_exists() && nzchar(avworkspace_name())) {
    ## e.g., from within AnVIL
    avworkflow_jobs() %>%
    ## select most recent workflow
    head(1) %>%
    ## find paths to output and log files on the bucket
    avworkflow_files()
}

if (gcloud_exists() && nzchar(avworkspace_name())) {
    avworkflow_localize(dry = TRUE)
}

avworkflow_configuration_template()

## Not run: 
config <-
    avworkflow_configuration("bioconductor-anvil-rpci", "AnVILBulkRNASeq")
str(config)

## End(Not run)

## Not run: 
avworkflow_import_configuration(config)

## End(Not run)