fastq to bam add provenance, was used in BSMN

fastq-to-bam-add-provenance.R

library(synapseClient)
synapseLogin()

table_id <- "syn7871084"
fileview_df <- synTableQuery(sprintf('select * from %s', table_id))@values

fastq2bam <- fileview_df %>%
    filter(fileFormat %in% c("fastq", "bam")) %>% # keep fastq and bam
    # group for cases where there are multiple files per file format
    group_by(sample_id_biorepository,
             sample_id_original, experiment_id, fileFormat) %>%
    # get all IDs by file format
    mutate(ids=paste(id, collapse=",")) %>%
    # get relevant fields, specifically excluding original id column
    select(sample_id_biorepository,
           sample_id_original, experiment_id, fileFormat, ids) %>%
    ungroup() %>%
    distinct() %>%
    # spread fastq and bam to separate columns
    spread(fileFormat, ids) %>%
    # keep those that are not null (have both bam and fastq)
    filter(!is.na(fastq), !is.na(bam))

addProvenance <- function(fastq, bam) {
    bams <- stringr::str_split(bam, ",")
    if (length(bams) > 1) {
        return(NULL)
    }
    else {
        ent <- synGet(bam, downloadFile=FALSE)
        generatedBy(ent) <- Activity(name="Alignment",
                                     description="fastq to bam alignment",
                                     used=stringr::str_split(fastq, ",")[[1]])
        return(generatedBy(synStore(ent, forceVersion=FALSE)))
    }
}