How to add tss_id and p_id in an Ensembl GTF (or any GTF other than that generated by cufflinks)

Question

[komal.rathi]

Hi everyone,

I am working with mm10 data and using the GRCm38 build 75 GTF from Ensembl. As everyone knows you need the tss_id and p_id to be present for differential isoform expression (by cuffdiff) when using any GTF other than the cufflinks' merged.gtf. I am using the following command to add the tss_id and p_id to my ensembl gtf:

cuffcompare \
  -o cuffcmp \
  -C -G \
  -r Mus_musculus.GRCm38.75.protein_linc.gtf \
  -s mm10.fa \
  Mus_musculus.GRCm38.75.protein_linc.gtf

To check whether I was doing it correctly, I checked the entries for a particular gene in both the input and output gtfs.

The 'gene' entry for Xkr4 in the original GTF looks like this:

chr1    protein_coding    gene    3205901    3671498    .    -    .    gene_id "ENSMUSG00000051951"; gene_name "Xkr4"; gene_source "ensembl_havana"; gene_biotype "protein_coding";

And, these are the entries corresponding to the above coordinates in the output GTF cuffcmp.combined.gtf:

chr1    processed_transcript    exon    3205901 3207317 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002060"; exon_number "1"; gene_name "Xkr4"; oId "ENSMUST00000162897"; nearest_ref "ENSMUST00000162897"; class_code "="; tss_id "TSS1356";
chr1    processed_transcript    exon    3213609 3216344 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002060"; exon_number "2"; gene_name "Xkr4"; oId "ENSMUST00000162897"; nearest_ref "ENSMUST00000162897"; class_code "="; tss_id "TSS1356";
chr1    processed_transcript    exon    3206523 3207317 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002061"; exon_number "1"; gene_name "Xkr4"; oId "ENSMUST00000159265"; nearest_ref "ENSMUST00000159265"; class_code "="; tss_id "TSS1357";
chr1    processed_transcript    exon    3213439 3215632 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002061"; exon_number "2"; gene_name "Xkr4"; oId "ENSMUST00000159265"; nearest_ref "ENSMUST00000159265"; class_code "="; tss_id "TSS1357";
chr1    protein_coding  exon    3214482 3216968 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002062"; exon_number "1"; gene_name "Xkr4"; oId "ENSMUST00000070533"; nearest_ref "ENSMUST00000070533"; class_code "="; tss_id "TSS1358"; p_id "P1235";
chr1    protein_coding  exon    3421702 3421901 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002062"; exon_number "2"; gene_name "Xkr4"; oId "ENSMUST00000070533"; nearest_ref "ENSMUST00000070533"; class_code "="; tss_id "TSS1358"; p_id "P1235";
chr1    protein_coding  exon    3670552 3671498 .       -       .       gene_id "XLOC_000653"; transcript_id "TCONS_00002062"; exon_number "3"; gene_name "Xkr4"; oId "ENSMUST00000070533"; nearest_ref "ENSMUST00000070533"; class_code "="; tss_id "TSS1358"; p_id "P1235";

In the output, the gene_id field has XLOC ids instead of Ensembl IDs. Can I fix this to have Ensembl IDs instead? Is there a better way to add tss_id and p_id to your Ensembl GTF?

Answer 1

[Malcolm.Cook]

I have developed an Rscript, cuffdiff_gtf_attributes, which can provided the additional attributes p_id and tss_id as required by cuffdiff to perform all the differential splicing/coding/expression contrasts. I have tested it with Ensembl GTF.

[Sukhi Singh]

Thanks for the script, works great!

Answer 2

[lavinia.gordon]

Hi,

You can download the correct gtf file from here I believe:

https://ccb.jhu.edu/software/tophat/igenomes.shtml