Fig. 1

Reverse cumulative distribution of the number of genes that have at least a given TPM. Both axes are logarithmic. Each line represents data from an individual SRA sample ID, quantified using the first iteration Kallisto transcriptome (i.e. a non-redundant set of Ensembl protein-coding CDS plus trimmed RefSeq mRNAs). Samples are not otherwise distinguished as in general, most relationships approximate the same power-law: a minority of genes account for the majority of reads. These relationships are piecewise linear because the capture of lowly expressed genes is noisy, an artefact of random transcriptome sampling. The vertical red line denotes TPM = 5. At higher values of TPM, the majority of samples have a log-linear relationship. Samples that deviate from this relationship are erroneous and excluded from subsequent analysis. Exponents of each sample’s log-log plot are given in Additional file 3: Table S3