For some analyses you might be interested in transforming the counts into TPMs which you can do with this function. This function uses the gene-level RPKMs to derive TPM values (see Details).
getTPM(rse, length_var = "bp_length", mapped_var = NULL)A RangedSummarizedExperiment-class object as downloaded with download_study.
A length 1 character vector with the column name from
rowData(rse) that has the coding length. For gene level objects
from recount this is bp_length. If NULL, then it will use
width(rowRanges(rse)) which should be used for exon RSEs.
A length 1 character vector with the column name from
colData(rse) that has the number of reads mapped. For recount RSE
object this would be mapped_read_count. If NULL (default)
then it will use the column sums of the counts matrix. The results are
different because not all mapped reads are mapped to exonic segments of the
genome.
A matrix with the TPM values.
For gene RSE objects, you will want to specify the length_var
because otherwise you will be adjusting for the total gene length instead
of the total exonic sequence length of the gene.
As noted in https://support.bioconductor.org/p/124265/, Sonali Arora et al computed TPMs in https://www.biorxiv.org/content/10.1101/445601v2 using the formula: TPM = FPKM / (sum of FPKM over all genes/transcripts) * 10^6
Arora et al mention in their code that the formula comes from
https://doi.org/10.1093/bioinformatics/btp692; specifically
1.1.1 Comparison to RPKM estimation where they mention an important
assumption: Under the assumption of uniformly distributed reads, we note
that RPKM measures are estimates of ...
There's also a blog post by Harold Pimentel explaining the relationship between FPKM and TPM: https://haroldpimentel.wordpress.com/2014/05/08/what-the-fpkm-a-review-rna-seq-expression-units/.
https://www.biorxiv.org/content/10.1101/445601v2 https://arxiv.org/abs/1104.3889
## Compute the TPM matrix from the raw gene-level base-pair counts.
tpm <- getTPM(rse_gene_SRP009615)