RPKM and TPM conversion?

[Jeremy]

 

Hi, I obtained some RNASeq processed data using RSEM in TACG. Unfortuniately I noticed that some data I downloaded use RPKM as the measurement for gene expression, and some other data I downloaded use TPM. Is it possible to convert RPKM into TPM? This two measurements are strongly correlated, but are not the same. In order to convert RPKM into TPM, what other information do I need to have? Thanks.

 

Jeremy

[Colin Dewey]

Hi Jeremy,

Assuming the RPKM values are calculated correctly (i.e., they are estimates of a quantity that is proportional to the relative abundance of each gene or transcript), you can convert to TPM simply by dividing each RPKM value by the sum of the RPKM values for all genes (or transcripts) and multiplying by one million. For further insight into what the conversion factor means, see section 1.1.1 of our paper:
http://bioinformatics.oxfordjournals.org/content/26/4/493.full
Briefly,
TPM = (mean transcript length in kilobases) x RPKM
where "mean transcript length" is the expression-weighted mean of the lengths of all isoforms. Because the mean transcript length can change from sample to sample, we have generally recommended the use of TPM instead of RPKM.

Best,
Colin

[b...@cs.wisc.edu]

Hi Jeremy,

In addition to Colin' s comment, you can also find description between the
relationship of RPKM and TPM at
http://deweylab.biostat.wisc.edu/rsem/rsem-calculate-expression.html .
Please see OUTPUT section sample_name.isoforms.results subsection.

Best,
Bo

[archana bhardwaj]

Hello everyone 

I am new to NGS . I have paired end sequence data for differential gene expression analysis. First i did the prepossing of my data such as adapter removal, ambiguous character and quality filtration . Now i want to assemble my data in form of contigs. I run trininty.pl to get the assembled contigs. There are so many output files . What is the exact contig files??? Inchworm contigs or bundled.fasta ??? How can i get the read count of each contig ??? 

waiting for reply 

[Anne Deslattes Mays]

Dear Colin,

Based upon your text -- shouldn't the formula read:
  TPM = (((mean transcript length in kilobases) x RPKM) / sum(RPKM all genes)) * 10^6.

Excellent paper btw -- very clear.  I used the method to convert other transcript counts measured by other instruments to TPMs -- but everyone can use the conversion from RPKMs (and FPKMs) to TPMs.

Any adjustment given FPKMs
Anne

[Anne Deslattes Mays]

One final adjustment -- length is reported not as kilobases, so the formula for transforming RPKM (and FPKM) to TPM is when transcript length is reported as bases rather than kilobases:

TPM = (((mean transcript length) x RPKM (or FPKM))/sum(RPKM (or FPKM) all genes)) * 10^3

[Colin Dewey]

Hi Anne,

Some corrections to your formulas:

If you have RPKM (single-end data) or FPKM (paired-end data) computed for a set of genes or transcripts you can convert to TPM with 

TPM = FPKM / (sum of FPKM over all genes/transcripts) * 10^6

The formula is the same for RPKM.

If you have read (single-end) or fragment (paired-end) counts, you can compute TPM by first computing RPKM (or FPKM) and then using the formula above for TPM.

An interesting aspect of the math here is that 

sum of FPKM over all transcripts = 10^6/(expression weighted mean transcript length in kb)

Best,

Colin

--
RSEM website: http://deweylab.biostat.wisc.edu/rsem/
---
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.