TPM, FPKM, or Normalized Counts? A Comparative Study of Quantification Measures for the Analysis of RNA-seq Data from the NCI Patient-Derived Models Repository

Abstract

Background: In order to correctly decode phenotypic information from RNA-sequencing (RNA-seq) data, careful selection of the RNA-seq quantification measure is critical for inter-sample comparisons and for downstream analyses, such as differential gene expression between two or more conditions. Several methods have been proposed and continue to be used. However, a consensus has not been reached regarding the best gene expression quantification method for RNA-seq data analysis.

Methods: In the present study, we used replicate samples from each of 20 patient-derived xenograft (PDX) models spanning 15 tumor types, for a total of 61 human tumor xenograft samples available through the NCI patient-derived model repository (PDMR). We compared the reproducibility across replicate samples based on TPM (transcripts per million), FPKM (fragments per kilobase of transcript per million fragments mapped), and normalized counts using coefficient of variation, intraclass correlation coefficient, and cluster analysis.

Results: Our results revealed that hierarchical clustering on normalized count data tended to group replicate samples from the same PDX model together more accurately than TPM and FPKM data. Furthermore, normalized count data were observed to have the lowest median coefficient of variation (CV), and highest intraclass correlation (ICC) values across all replicate samples from the same model and for the same gene across all PDX models compared to TPM and FPKM data.

Conclusion: We provided compelling evidence for a preferred quantification measure to conduct downstream analyses of PDX RNA-seq data. To our knowledge, this is the first comparative study of RNA-seq data quantification measures conducted on PDX models, which are known to be inherently more variable than cell line models. Our findings are consistent with what others have shown for human tumors and cell lines and add further support to the thesis that normalized counts are the best choice for the analysis of RNA-seq data across samples.

Keywords: Count; DESeq2; FPKM; Normalization; Patient derived xenograft models; Quantification measures; RNA sequencing; RSEM; TMM; TPM.

Fig. 1

A Hierarchical clustering of 61 patient-derived xenograft (PDX) samples using TPM data. B Hierarchical clustering of 61 PDX samples using DESeq2 normalized count data. Distance metric 1-Pearson correlation was used to generate the dendrogram in each right panel and Euclidean distance was used for the dendrogram in each left panel. Discordant models are highlighted with different color labels

Fig. 2

Bar plot of median coefficients of variation (CV) for gene expression levels from replicate samples of each PDX model using different quantification measures

Fig. 3

A Bar plot of gene intraclass correlation coefficients (ICC_g) across replicate samples of each PDX model using different quantification measures. B Boxplots of model intraclass correlation coefficients (ICC_m) for gene expression levels from replicate samples across 20 PDX models using different quantification measures

Fig. 4

A Pairwise scatter plots comparing TPM values for all genes between replicate samples of PDX model 475296-252-R. B Pairwise scatter plots comparing DESeq2 normalized count values for all genes between replicate samples of PDX model 475296-252-R. The x- and y- axes are normalized log₂ counts on all pairwise scatter plots. Plots along the diagonal represent the density of the respective variable

Fig. 5

A Bar plot of the sum of TPM values for the top 5 most highly expressed genes in four PDX models with the lowest ICC_g. B Bar plot of the sum of TPM values for the top 5 most highly expressed genes in five PDX models with the highest ICC_g