In mpkocher/smrtlink-system-stats:

r params$smrtlink_system System DataSet Summary

Load all DataSet types

library(dplyr)
library(scales)
library(ggplot2)
library(lubridate)
library(jsonlite)
library(tibble)
library(purrr)
library(tidyverse)
library(stringr)
library(readr)
library(forcats)
library(smrtlinkr)

system_name <- params$smrtlink_system
created_at <- params$created_at
root_data_dir <- params$root_data_dir

summary_msg <- paste0("Created at ", created_at," with SMRT Link system ", system_name)
print(summary_msg)

smrtlink_status <- load_status_from_json_path(to_status_json_path(root_data_dir, system_name))
print(smrtlink_status)

to_system_ds_json_path <- function(ds_shortname) { return(to_ds_json_path(root_data_dir, system_name, ds_shortname))}

# create these manually to refer to them later
path_referenceset <-to_system_ds_json_path("references")
path_subreadset <- to_system_ds_json_path("subreads") 
path_ccsreads <- to_system_ds_json_path("ccsreads")
path_alignmentset <- to_system_ds_json_path("alignments")
path_barcodeset <- to_system_ds_json_path("barcodes")
path_ccsalignments <- to_system_ds_json_path("ccsalignments")
path_contigs <- to_system_ds_json_path("contigs")
path_gmap_ref <- to_system_ds_json_path("gmapreferences")
path_hdfsubreads <- to_system_ds_json_path("hdfsubreads")


ds_paths <-
  c(
  path_referenceset,
  path_subreadset,
  path_ccsreads,
  path_alignmentset,
  path_barcodeset,
  path_ccsalignments,
  path_contigs,
  path_gmap_ref,
  path_hdfsubreads
  )

tdf <- load_dataset_metadata_from_json_paths(ds_paths)

System DataSet Summary

# Raw summary. There are many values that are negative due to bugs, or default values of -1
tdf %>% to_summary_dataset() %>% print()

System DataSet Summary By DataSet Type

tdf %>% group_by(dataset_metatype) %>% summarize(total=n()) %>% print()

System Summary DataSet By DataSet Version

tdf %>% group_by(version) %>% summarize(total=n()) %>% print()

SubreadSet Full Summary Details

subread_df <- load_dataset_metadata_from_json_path(path_subreadset)
to_full_summary_dataset(subread_df, "SubreadSet")

ReferenceSet Full Summary Details

ref_df <- load_dataset_metadata_from_json_path(path_referenceset)
to_full_summary_dataset(ref_df, "ReferenceSet")

GmapReferenceSet Full Summary

# allow a little slop when there are no datasets
gmap_df <- load_dataset_metadata_from_json_path(path_gmap_ref)
to_full_summary_dataset(gmap_df, "GmapReferenceSet")

ContigSet Full Summary

contig_df <- load_dataset_metadata_from_json_path(path_contigs)
to_full_summary_dataset(contig_df, "ContigSet")

CCS Set Full Summary

ccs_df <- load_dataset_metadata_from_json_path(path_ccsreads)
to_full_summary_dataset(ccs_df, "CCSSet")

Alignment Set Full Summary

align_df <- load_dataset_metadata_from_json_path(path_alignmentset)
to_full_summary_dataset(align_df, "AlignmentSet")

CCS AlignmentSet Full Summary

ccs_align_df <- load_dataset_metadata_from_json_path(path_ccsalignments)
to_full_summary_dataset(ccs_align_df, "CCS AlignmentSet")

HdfSubreadSet Full Summary

hdf_df <- load_dataset_metadata_from_json_path(path_hdfsubreads)
to_full_summary_dataset(hdf_df, "HdfSubreadSet")

mpkocher/smrtlink-system-stats documentation built on May 29, 2019, 5:45 a.m.