compute_scale_factors: Compute count scaling factors
In LieberInstitute/recount3: Explore and download data from the recount3 project

compute_scale_factors

R Documentation

Compute count scaling factors

Description

This function computes the count scaling factors used by transform_counts(). This function is similar to recount::scale_counts(factor_only = TRUE), but it is more general.

Usage

compute_scale_factors(
  x,
  by = c("auc", "mapped_reads"),
  targetSize = 4e+07,
  L = 100,
  auc = "recount_qc.bc_auc.all_reads_all_bases",
  avg_mapped_read_length = "recount_qc.star.average_mapped_length",
  mapped_reads = "recount_qc.star.all_mapped_reads",
  paired_end = is_paired_end(x, avg_mapped_read_length)
)

Arguments

`x`	Either a RangedSummarizedExperiment-class created by `create_rse()` or the sample metadata created by `read_metadata()`.
`by`	Either `auc` or `mapped_reads`. If set to `auc` it will compute the scaling factor by the total coverage of the sample. That is, the area under the curve (AUC) of the coverage. If set to `mapped_reads` it will scale the counts by the number of mapped reads (in the QC annotation), whether the library was paired-end or not, and the desired read length (`L`).
`targetSize`	A `numeric(1)` specifying the target library size in number of single end reads.
`L`	A `integer(1)` specifying the target read length. It is only used when `by = 'mapped_reads'` since it cancels out in the calculation when using `by = 'auc'`.
`auc`	A `character(1)` specifying the metadata column name that contains the area under the coverage (AUC). Note that there are several possible AUC columns provided in the sample metadata generated by `create_rse()`.
`avg_mapped_read_length`	A `character(1)` specifying the metdata column name that contains the average fragment length after aligning. This is typically twice the average read length for paired-end reads.
`mapped_reads`	A `character(1)` specifying the metadata column name that contains the number of mapped reads.
`paired_end`	A `logical()` vector specifying whether each sample is paired-end or not.

Value

A numeric() with the sample scale factors that are used by transform_counts().

Examples


## Download the metadata for SRP009615, a single-end study
SRP009615_meta <- read_metadata(
    metadata_files = file_retrieve(
        locate_url(
            "SRP009615",
            "data_sources/sra",
        )
    )
)

## Compute the scaling factors
compute_scale_factors(SRP009615_meta, by = "auc")
compute_scale_factors(SRP009615_meta, by = "mapped_reads")

## Download the metadata for DRP000499, a paired-end study
DRP000499_meta <- read_metadata(
    metadata_files = file_retrieve(
        locate_url(
            "DRP000499",
            "data_sources/sra",
        )
    )
)

## Compute the scaling factors
compute_scale_factors(DRP000499_meta, by = "auc")
compute_scale_factors(DRP000499_meta, by = "mapped_reads")

## You can compare the factors against those from recount::scale_counts()
## from the recount2 project which used a different RNA-seq aligner
## If needed, install recount, the R/Bioconductor package for recount2:
# BiocManager::install("recount")
recount2_factors <- recount::scale_counts(
    recount::rse_gene_SRP009615,
    by = "auc", factor_only = TRUE
)
recount3_factors <- compute_scale_factors(SRP009615_meta, by = "auc")
recount_factors <- data.frame(
    recount2 = recount2_factors[order(names(recount2_factors))],
    recount3 = recount3_factors[order(names(recount3_factors))]
)
plot(recount2 ~ recount3, data = recount_factors)
abline(a = 0, b = 1, col = "purple", lwd = 2, lty = 2)

LieberInstitute/recount3 documentation built on May 4, 2024, 4:16 a.m.