TPM or FPKM questions in TCGA data

[Tsung-Ching Lai]

Hi everyone

  I just compared TCGA LUAD's data by re-calculating the results by RSEM, tophat-cufflinks and TCGA rawdata.

  In the below table, the expression trends are similar among RSEM and tophat-cufflinks but not in the last column TCGA-62-8402-01.

  I am afraid that my command is wrong. Did any one can give me some suggestion or the problems?

#  My system environment
    iGenome-Ensembl-human-GRCh37- gene.gtf and genome.fa (Bowtie2)
    Bowtie2-2.1.0

    RSEM-1.2.18 :

    rsem-prepare-reference --gtf ./genes.gtf --bowtie2 --bowtie2-path /home/"myID"/bowtie2-2.1.0/ ./Bowtie2Index/genome.fa ./rsem_ref/hg19

    rsem-calculate-expression --bowtie2-path /home/peter/Downloads/bowtie2-2.1.0 --paired-end -p 6 ./*_1.fastq ./*_2.fastq /home/"myID"/rsem_ref/hg19 TCGA0002_rsem

    Tophat-2.0.11 + cufflinks-2.1.1

    tophat -p 8 -G /work/"myID"/genes.gtf --no-coverage-search /work/"myID"/Bowtie2_Index/genome ./*_1.fastq ./*_2.fastq; rm ./*.fastq; cufflinks -p 8 -G ../genes.gtf -b ../Bowtie2_Index/genome.fa ./tophat_out/accepted_hits.bam


The results in the same sample

gene_id	gene_short_name	locus	FPKM-cufflinks	TPM-rsem	FPKM-rsem	TCGA-62-8402-01
ENSG00000000005	TNMD	X:99839798-99854882	0	0	0	0
ENSG00000001561	ENPP4	6:46097700-46114435	2.76868	4.61	2.96	8.2385
ENSG00000005073	HOXA11	7:27221128-27224842	2.21098	2.7	1.74	6.4946
ENSG00000005075	POLR2J	7:102113564-102119354	55.197	105.06	67.52	9.9242
ENSG00000005513	SOX8	16:1031807-1036979	0.0603328	0.1	0.07	2.442
ENSG00000006059	KRT33A	17:39502343-39507064	0	0	0	0
ENSG00000006075	CCL3	17:34415601-34417515	9.08015	36.39	23.39	8.4706
ENSG00000006116	CACNG3	16:24266873-24374122	0	0	0	0
ENSG00000006128	TAC1	7:97361219-97369784	0	0	0	0
ENSG00000006606	CCL26	7:75398850-75419214	0.352621	0.46	0.3	1.984
ENSG00000006788	MYH13	17:10201400-10276447	0	0	0	0
ENSG00000008438	PGLYRP1	19:46522410-46526323	0.0367952	0.05	0.03	0.5779

By the way, is TPM more suitable for TCGA comparison among all samples? I read Dr. Patcher's blog (https://liorpachter.wordpress.com/tag/fpkm/), according to the formula of FPKM the value would not be consist due to the size of sequencing reads pool. However, I read in TCGA forum that most TCGA value in RNA-seq is FPKM or RPKM. Which data is more reliable??

Thank you very much.

Best Regards,
Chuching

[Bo Li]

Hi Chuching,

If you want to compare the *** relative *** expression values of a same
gene across different samples, you should use TPM. It is very easy to
convert FPKM/RPKMs to TPMs. You first normalize your FPKMs from all
genes so that their sum is equal to 1. Then you product 1e6 to each
normalized value and you will obtain TPMs. If you want to compare
absolute expression across samples, you need to do within-sample
normalization, such as TMM (edgeR) and median normalization (DESeq).

For your first question, since I am not involved in analyzing TCGA data,
I'm not sure.

Colin, do you have any suggestions?

Hope it helps,
Bo

> --
> RSEM website: http://deweylab.biostat.wisc.edu/rsem/ [1]
> ---
> You received this message because you are subscribed to the Google
> Groups "RSEM Users" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to rsem-users+...@googlegroups.com.
> To post to this group, send email to rsem-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/rsem-users [2].
>
>
> Links:
> ------
> [1] http://deweylab.biostat.wisc.edu/rsem/
> [2] http://groups.google.com/group/rsem-users