oc2bioc – notes for developers

Vince Carey^[

Source: vignettes/devel.Rmd

Interacting with the open-cravat infrastructure

The oc2bioc package (currently at github.com/vjcitn/oc2bioc) incorporates the open-cravat python modules. In the main user vignette we invoke populate_module_set, to learn about available modules in the OpenCRAVAT system. The oc2bioc package will grow to include formal interfaces to other key feature of OpenCRAVAT.

Formal interfacing using basilisk

Let’s have a look at populate_module_set:

library(oc2bioc)
#> Loading required package: reticulate
#> Loading required package: httr
populate_module_set
#> function () 
#> {
#>     proc = basilisk::basiliskStart(ocbenv)
#>     on.exit(basilisk::basiliskStop(proc))
#>     basilisk::basiliskRun(proc, function() {
#>         cra = reticulate::import("cravat", convert = TRUE)
#>         modnames = cra$admin_util$search_remote(".*")
#>         allinfo = lapply(modnames, function(x) cra$admin_util$get_remote_module_info(x))
#>         atts = names(allinfo[[1]])
#>         vals = lapply(allinfo, function(x) {
#>             tmp = lapply(atts, function(y) x[[y]])
#>             names(tmp) = atts
#>             do.call("new", c("OpenCRAVATModule", tmp))
#>         })
#>         names(vals) = modnames
#>         new("OpenCRAVATModuleSet", modset = vals, created = date())
#>     })
#> }
#> <bytecode: 0x5589a9d7c9b0>
#> <environment: namespace:oc2bioc>

While this structure may appear complex, it is schematic and easily thought of as a template. The purpose of the basilisk package functions is isolation of the open-cravat code and its dependencies to a mutually consistent environment.

Informal view of capabilities of the python stack using reticulate

In order to understand the features of the open-cravat python stack, we can use reticulate on any system in which open-cravat is installed and accessible via PYTHONPATH.

Below we see how to use reticulate to interrogate the cravat module and its submodules from R. There is a lot of software in this stack. Choosing functions to expose in R/Bioconductor is a long-term project to be driven by user interest.

> library(reticulate)
1/5 packages newly attached/loaded, see sessionInfo() for details.
> cr = import("cravat")
> names(cr)
 [1] "admin_util"             "aggregator"             "BadFormatError"        
 [4] "base_annotator"         "base_commonmodule"      "base_converter"        
 [7] "base_mapper"            "base_postaggregator"    "BaseAnnotator"         
[10] "BaseCommonModule"       "BaseConverter"          "BaseMapper"            
[13] "BasePostAggregator"     "config_loader"          "ConfigLoader"          
[16] "ConfigurationError"     "constants"              "Cravat"                
[19] "cravat_class"           "cravat_filter"          "cravat_report"         
[22] "cravat_util"            "CravatFilter"           "CravatProfileException"
[25] "CravatReport"           "crx_def"                "exceptions"            
[28] "ExpectedException"      "FileIntegrityError"     "get_live_annotator"    
[31] "get_live_mapper"        "get_module"             "get_ucsc_bins"         
[34] "get_wgs_reader"         "inout"                  "InvalidData"           
[37] "InvalidReporter"        "KillInstallException"   "LiftoverFailure"       
[40] "LiveAnnotator"          "mp_runners"             "raise_break"           
[43] "reverse_complement"     "run"                    "run_reporter"          
[46] "signal"                 "store_utils"            "switch_strand"         
[49] "translate_codon"        "util"                   "wgs"                   
> cr$admin_util
Module(cravat.admin_util)
> head(names(cr$admin_util))
[1] "change_password" "check_login"     "compare_version" "constants"      
[5] "copy"            "create_account"

Interacting with the REST API for local deployments

See the documentation for queryOC in the oc2bioc package for information on how RESTful queries can be resolved with locally installed annotators.