which Length of gene is valid for using in Transcript Per million (TPM)?

Question

[modarzi]

Hi. for calculating Transcript Per Milone (TPM) from TCGA HTseq-count I need gene length. also, I used gene code V.22 for annotation which has different columns for each gene. I bring one record from annotation file as an example:

feature     start       end  score  strand  frame   gene_id         gene_name   
 gene       3281801   32897826  .     +       .   ENSG00000206557.5    TRIM71   

full_length   exon_length    exon_num       first_exon            last_exon             
    79809          8685            4      ENSE00001538095.1 ENSE00001498538.5   
one_transcript        one_transcript_start      one_transcript_end 
ENST00000383763.5            32818018                32897826

As you see, for each ensemble gene, we have full_length and exon_length. Now, for TPM calculating I need to 'gene length'. please guide me on which length should I use for TPM?

[GenoMax]

Prior threads for consideration:

Calculating TPM Values
Calculating TPM from featureCounts output

Answer 1

[Kevin Blighe]

Hello again,

Assuming that you have read counts per gene (not per exon), please use full_length.

Kevin

[modarzi]

Thanks. Dear Dr. Blighe, I have another question relate to this problem. as you said, I have to use 'full_lenght' for calculating TPM. So, in my example, 'full_lenght' of 'TRIM71' is 79809. I appreciate it if you guide me should I use 79809 or 79.809? in other words, I have to use gene length based on bp or kbp? and 79809 is bp or kbp? Best Regards

[Kevin Blighe]

Hi, you can use bp; so, 79809.