Formatting salmon output to match GTF

[comprna]

Formatting salmon output to match GTF

[EduEyras]

Hi, The idea is simply to make sure that the IDs are the same. The error must be because your input is not the expected format. Can you send us a sample of your expression file? Thanks E Dr. Eyras ICREA Research Professor Universitat Pompeu Fabra Barcelona Spain

…

On 26 Jul 2018, at 01:27, ibmuller ***@***.***> wrote: In the tutorial there is this part: For the calculation of the events, SUPPA2 requires a GTF file. Here we provide one from Ensembl. The transcripts ids in our iso_tpm file should be equal as the ids in the GTF. We have a small script in R for formatting this ids ~/Rscript ~/scripts/format_Ensembl_ids.R iso_tpm.txt Here is the file already formatted with the TPM of the isoforms. The R script would then change the the transcript ids to match the GTF file so I tried it on my own salmon output but it gives me an error saying this: Rscript ensembl_ids.R ~/data/suppa_output/cemwt_all_tpm.txt Error in `row.names<-.data.frame`(`*tmp*`, value = value) : duplicate 'row.names' are not allowed Calls: rownames<- -> row.names<- -> row.names<-.data.frame In addition: Warning message: non-unique values when setting 'row.names': Execution halted Is there another way to change the file to match the GTF? im not even sure what the R script does exactly. regards, I. Muller — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[ibmuller]

you mean the FASTQ file i used for the quantification? i made a sample of one of the replicates using seqtk:
sample_10k.zip

I used the provided Ensembl hg19 to create an salmon index and then used the standard salmon quant command on the fastq file to create the quant.sf file. It is a stranded reverse library type.

I didnt get any error messages throughout this, only with the R script.

Hope this helps.

[pent1162]

I also have the same problem. I use hg38. The iso_tpm.txt is totally different from hg19.

[EduEyras]

you mean the FASTQ file i used for the quantification? i made a sample of one of the replicates using seqtk:
sample_10k.zip

I used the provided Ensembl hg19 to create an salmon index and then used the standard salmon quant command on the fastq file to create the quant.sf file. It is a stranded reverse library type.

I didnt get any error messages throughout this, only with the R script.

Hope this helps.

Hi, you were using the Salmon output directly. The required input is just a list of transcript IDs and TPM values. Perhaps your file had the TPMs in the wrong columns?

E.

[EduEyras]

I also have the same problem. I use hg38. The iso_tpm.txt is totally different from hg19.

The GTF you use to calculate the events should match the transcript IDs you used
for quantification.

If you happen to have the transcript quantification from an annotation on hg38, but the GTF from hg19, it might work if you trust that transcript sequences have not changed (or not much).

You can check that looking at the transcript ID and version: https://www.ensembl.org/info/genome/stable_ids/index.html

I hope this helps

E.

[pent1162]

hg38_iso_tpm.txt
hg19_iso_tpm.txt
I used Homo_sapiens.GRCh38.87.gtf as the annotation.
hg19_iso_tpm.txt, the first column is different from hg38_iso_tpm.
If I used hg38_iso_tpm.txt run "~/Rscript ~/scripts/format_Ensembl_ids.R iso_tpm.txt"
I got the same error.
If I jump to suppy.py psiPerEvent, I got lots of error:
ERROR:psiCalculator:PSI not calculated for event ENSG00000100429;SE:22:50250160-50250427:50250523-50250771:-.
ERROR:psiCalculator:transcript ENST00000489424 not found in the "expression file".
then the output are all NA.
how could I solve the problem?
Does GRCh38.87.gtf need to run through the format_Ensembl_ids.R first?
Thanks

[EduEyras]

Hi, The error is because the GTF produced events involving transcripts that are not in the expression file. If you still got some PSIs calculated it means that some transcript IDs are still maintained. Do you have the means to calculate the TPM values for the transcripts in GRCh38.87? That would solve the problem. Sometimes it also happens that the GTF contains annotations for non-standard chromosomes (other than chr1,...,chr22, chrX, chrY) but the quantification is only available for transcripts annotated in those standard chromosomes. This will throw the same error, but should not be a problem if you don't care about events in non-standard chromosomes. Let me know if this helps cheers E.

…

On Wed, Oct 10, 2018 at 5:36 AM PVenRT ***@***.***> wrote: hg38_iso_tpm.txt <https://github.com/comprna/SUPPA/files/2463019/hg38_iso_tpm.txt> hg19_iso_tpm.txt <https://github.com/comprna/SUPPA/files/2463020/hg19_iso_tpm.txt> I used Homo_sapiens.GRCh38.87.gtf as the annotation. hg19_iso_tpm.txt, the first column is different from hg38_iso_tpm. If I used hg38_iso_tpm.txt run "~/Rscript ~/scripts/format_Ensembl_ids.R iso_tpm.txt" I got the same error. If I jump to suppy.py psiPerEvent, I got lots of error: ERROR:psiCalculator:PSI not calculated for event ENSG00000100429;SE:22:50250160-50250427:50250523-50250771:-. ERROR:psiCalculator:transcript ENST00000489424 not found in the "expression file". then the output are all NA. how could I solve the problem? Does GRCh38.87.gtf need to run through the format_Ensembl_ids.R first? Thanks — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#37 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AMWVBzhZaZX1ZcRlREt4r95S8FKiR6tTks5ujWtPgaJpZM4VgTOb> .

-- Dr E Eyras ICREA Research Professor Universitat Pompeu Fabra PRBB, Dr Aiguader 88 Tel: [phone] E08003 Barcelona, Spain Fax: [phone] http://scholar.google.com/citations?user=LiojlGoAAAAJ http://www.researcherid.com/rid/L-1053-2014 http://regulatorygenomics.upf.edu/

[pent1162]

I solved this problem by formatting the gtf and my iso_tpm files to what you provided.
Here is the result:

btw, I followed your tutorial. In the volcano plotting step, I can't find "TRA2_diffSplice_avglogtpm.tab" so this script can't work. Did I miss any step in this tutorial?
Thanks.

[EduEyras]

Hi,

thanks for the note about the .tab file. We will include the script to generate this file.

Standard GTF should work. Were you using a different GTF format? It'll be good to know if some form of GTF does not work well.

Thanks

Eduardo

[JLTrincado]

Hi,

The file "TRA2_diffSplice_avglogtpm.tab" is generated activating the flag --save_tpm_events. It generates a file in the output folder with the average log tpm of the events. I have modified the associated script.

Thanks for your interest in our software.
Juanlu.

[pent1162]

I use GRCH38.87, downloading from ensembl.
I think it just need to reformat the gtf and isoform output due ENST contain ".x" version.
btw, can I get the raw number of inclusion & exclusion read for each event from SUPPA2 and how?

[EduEyras]

Hi, If your quantification is based on the same IDs you shouldn't need to modify the GTF. However if the transcript quantification is done with IDs without the version, yes you need to modify the file. However, if they are not exactly the same version I would double check that this does not lead to possible inconsistencies or artefacts. With SUPPA you cannot go back to the raw reads. That's the whole point of the method. On the other hand, if you can get a pseudoBAM from Salmon or Kallisto, that will pribably tell you how many reads fall on the transcrips that contribute to either form of the local splicing event. These are not the classical local counts falling on junctions, etc, but will give you an idea of the coverage for that event. If you rely on counting reads mapped to the genome in the classical way, there may be discrepancies with the assignments made by Salmon or kallisto. I hope this helps Best Eduardo Dr. Eyras ICREA Research Professor Universitat Pompeu Fabra Barcelona Spain

…

On 25 Oct 2018, at 08:15, PVenRT ***@***.***> wrote: I use GRCH38.87, downloading from ensembl. I think it just need to reformat the gtf and isoform output due ENST contain ".x" version. btw, can I get the raw number of inclusion & exclusion read for each event from SUPPA2 and how? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.