inst/scripts/make-data_HumanLungCancerV3.R

In julia-wrobel/VectraPolarisData: Vectra Polaris and Vectra 3 multiplex single-cell imaging data

# Human Lung Cancer from Vectra 3 experiment
library(tidyverse)
library(SpatialExperiment)
library(VectraPolarisData)

source("inst/scripts/make-data.R")

######################################################################
# Convert Inform-processed mIHC data to SpatialExperiment object
sample_path = "data"
spe_lung = readVectraTable(sample_path = sample_path,
                           save = FALSE)
# Data has 8 markers for 1604786 cells.
# 153 subjects
# 761 total images

######################################################################
# Load clinical data
file_path_clinical = "data/Lung_clinical.csv"
clinical_data = read_csv(file_path_clinical)

######################################################################
# Process clinical data

# survival_status is censoring variable 0 = alive, 1 = dead
clinical_data = clinical_data %>%
  janitor::clean_names() %>%
  # create mhcII high/low variable
  mutate(mhcII_status = ifelse(floor(ck_hlad_rpos_total) > 5, "high", "low")) %>%
  # select variables of interest
  select(patient_id, gender, mhcII_status, age_at_diagnosis, stage_at_diagnosis, stage_numeric, pack_years,
         survival_days, survival_status = censor_0_alive_1_dead,
         #t, n, m, largest_size, number_ln_taken, number_ln, lvi,  # commenting out because don't know what these variables are
         #therapy_type, number_cycles, # these don't seem too useful from a data analysis standpoint (lots of NAs)
         cause_of_death, adjuvant_therapy,
         time_to_recurrence_days, recurrence_or_lung_ca_death)


######################################################################
# Add clinical data to spe object via colData

# process patient id variable (called slide_id in the Vectra data) to match up with Vectra data
vectra_id <- tibble(slide_id = unique(colData(spe_lung)$slide_id),
       id = slide_id) %>%
  separate(id, into = c("number", "patient_id"), sep = " ") %>%
  select(-number)

# merge with clinical_data
clinical_data <- inner_join(vectra_id, clinical_data, by = "patient_id") %>%
  select(-patient_id) %>% as.data.frame()

# add as table to the dataset as metadata
metadata(spe_lung)$clinical_data <- clinical_data

######################################################################
# save spe object
saveRDS(spe_lung, file = "data/spe_lung.rds")