V17 OpenPBTA Histologies File QC

Load packages

suppressMessages(library(emo))

Directories and Files

Directories

# Data directory
new_data_dir <- file.path("~/Documents/GitHub/jharenza.github.io/openPBTA-notebooks/V17-histologies-QC/data/")
pbta_data_dir <- file.path("~/OpenPBTA-analysis/data/release-v16-20200320/")

Input Filepaths

pbta_hist_file <- file.path(pbta_data_dir, "pbta-histologies.tsv")
v17_hist_file <- file.path(new_data_dir, "Open PBTA v17 Candidate - v2.tsv")

Read in files

v16_hist <- read.delim(pbta_hist_file, sep = "\t", header = T)
v17_hist <- read.delim(v17_hist_file, sep = "\t", header = T)

First, assess dimensions whether new column names match the old

New columns: pathology_free_text_diagnosis and cohort_participant_id and glioma_brain_region -> CNS_region ✅

dim(v16_hist)

[1] 2833   33

dim(v17_hist)

[1] 2833   35

setdiff(names(v16_hist), names(v17_hist))

[1] "glioma_brain_region"

setdiff(names(v17_hist), names(v16_hist))

[1] "CNS_region"                    "pathology_free_text_diagnosis" "cohort_participant_id"        

Check levels of histology columns

These levels look as expected ✅

as.data.frame(table(v17_hist$pathology_diagnosis))

as.data.frame(table(v17_hist$integrated_diagnosis))

as.data.frame(table(v17_hist$broad_histology))

as.data.frame(table(v17_hist$short_histology))

Check tables from v16 -> v17 for fields not expected to change (quick overview)

Experimental strategy ✅

table(v16_hist$experimental_strategy)

  Panel RNA-Seq     WGS     WXS 
      2    1028    1741      62 

table(v17_hist$experimental_strategy)

  Panel RNA-Seq     WGS     WXS 
      2    1028    1741      62 

Sample Type ✅

table(v16_hist$sample_type)

Normal  Tumor 
   833   2000 

table(v17_hist$sample_type)

Normal  Tumor 
   833   2000 

Composition ✅

as.data.frame(table(v16_hist$composition))

as.data.frame(table(v17_hist$composition))

Broad composition ✅

as.data.frame(table(v16_hist$broad_composition))

as.data.frame(table(v17_hist$broad_composition))

Germline sex estimate ✅

as.data.frame(table(v16_hist$germline_sex_estimate))

as.data.frame(table(v17_hist$germline_sex_estimate))

RNA library ✅

as.data.frame(table(v16_hist$RNA_library))

as.data.frame(table(v17_hist$RNA_library))

Cohort ✅

as.data.frame(table(v16_hist$cohort))

as.data.frame(table(v17_hist$cohort))

Sample and aliquot IDs ❌ - Aliquot ID for PNOC003 BS_MVYA262V sample 7316-3213 was changed from A05265 to A05233. Is this correct? ❗

setdiff(v16_hist[, c("Kids_First_Biospecimen_ID", "sample_id", "aliquot_id", "cohort")],
        v17_hist[, c("Kids_First_Biospecimen_ID", "sample_id", "aliquot_id", "cohort")])

setdiff(v17_hist[, c("Kids_First_Biospecimen_ID", "sample_id", "aliquot_id", "cohort")],
        v16_hist[, c("Kids_First_Biospecimen_ID", "sample_id", "aliquot_id", "cohort")])

Sequencing Center - looks like BS_QB84TBA3 and BS_S2TA8R29 previously NA were updated to NantOmics. Is this correct? ❗

as.data.frame(table(v16_hist$seq_center))

as.data.frame(table(v17_hist$seq_center))

setdiff(v16_hist[, c("Kids_First_Biospecimen_ID", "seq_center")],
        v17_hist[, c("Kids_First_Biospecimen_ID", "seq_center")])

setdiff(v17_hist[, c("Kids_First_Biospecimen_ID", "seq_center")],
        v16_hist[, c("Kids_First_Biospecimen_ID", "seq_center")])

Tumor/Normal Fraction and Ploidy ✅

setdiff(v16_hist[, c("Kids_First_Biospecimen_ID", "normal_fraction", "tumor_fraction", "tumor_ploidy")],
        v17_hist[, c("Kids_First_Biospecimen_ID", "normal_fraction", "tumor_fraction", "tumor_ploidy")])

setdiff(v17_hist[, c("Kids_First_Biospecimen_ID", "normal_fraction", "tumor_fraction", "tumor_ploidy")],
        v16_hist[, c("Kids_First_Biospecimen_ID", "normal_fraction", "tumor_fraction", "tumor_ploidy")]) 

Tumor Descriptor ❌

as.data.frame(table(v16_hist$tumor_descriptor))

sum(as.data.frame(table(v16_hist$tumor_descriptor))$Freq)

[1] 2000

as.data.frame(table(v17_hist$tumor_descriptor))

sum(as.data.frame(table(v17_hist$tumor_descriptor))$Freq)

[1] 2000

Which are the samples that have had Tumor Descriptor changes?

105 samples, mostly CBTTC - Seems like a lot, but could be reasonable with the refresh. Is this correct? ❗

setdiff(v17_hist[,c("Kids_First_Biospecimen_ID", "tumor_descriptor", "cohort")], v16_hist[,c("Kids_First_Biospecimen_ID", "tumor_descriptor", "cohort")])

Did new subtypes get captured?

Went from 493 to 1015 subtypes, BRAF subtypes look to be updated, LGG subtypes added. ✅

as.data.frame(table(v16_hist$molecular_subtype))

sum(as.data.frame(table(v16_hist$molecular_subtype))$Freq)

[1] 493

as.data.frame(table(v17_hist$molecular_subtype))

sum(as.data.frame(table(v17_hist$molecular_subtype))$Freq)

[1] 1015

How many biospecimen pathology diagnoses were changed?

33

setdiff(v17_hist[,c("Kids_First_Biospecimen_ID", "pathology_diagnosis", "cohort")], v16_hist[,c("Kids_First_Biospecimen_ID", "pathology_diagnosis", "cohort")])

Column `pathology_diagnosis` joining factors with different levels, coercing to character vector

# Add to df
path_diff <- as.data.frame(setdiff(v17_hist[,c("Kids_First_Biospecimen_ID", "pathology_diagnosis", "cohort")], v16_hist[,c("Kids_First_Biospecimen_ID", "pathology_diagnosis", "cohort")]))

Column `pathology_diagnosis` joining factors with different levels, coercing to character vector

# How many per cohort?
table(path_diff$cohort)

  CBTTC PNOC003 
     12      21 

Are the integrated_diagnosis, broad_histology, and short_histology NA for these samples? Looks like yes for CBTTC, but no for PNOC0003 ❗

v17_hist[v17_hist$Kids_First_Biospecimen_ID %in% path_diff$Kids_First_Biospecimen_ID,c("Kids_First_Biospecimen_ID", "pathology_diagnosis", "integrated_diagnosis", "broad_histology", "short_histology", "cohort")]

Note: For consistency of the process, we could make these NA for the PNOC003 samples for which pathology_diagnosis changed, however, the argument against this could be that we know all PNOC003 samples are DMGs and thus would not have integrated_diagnosis, broad_histology, and short_histology changes, so keeping these values and subtyping them from those values would be fine.