Hi,

This question is in reference to the recently updated TxImport vignette: https://bioconductor.org/packages/3.7/bioc/vignettes/tximport/inst/doc/tximport.html 

I saw a related question from a few months back but wasn't able to follow the solution and am posting again, thanks in advance.

I ran Salmon on my fastq files from a recent mouse RNAseq experiment using the Transcript Sequences CHR Fasta file from the current Gencode Mouse release (https://www.gencodegenes.org/releases/current.html / ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_human/release_27/gencode.v27.transcripts.fa.gz). Everything seemed to go smoothly and so now I am in the process of making a tx2gene table to do my feature counting with tximport.

I note in your example that you import a constructed tx2gene .csv table from a recent gencode release. 

library(readr)
tx2gene <- read_csv(file.path(dir, "tx2gene.gencode.v27.csv"))
head(tx2gene)

I'm trying to get to this point with the gencode.vM16.annotation.gtf file I downloaded from the most recent release, but am not completely following how to go about this. Would you mind spelling this out for me a bit more? This is what I've done and want to make sure I haven't missed anything:

> TxDb <- makeTxDbFromGFF(file = "/Path/to/file/gencode.vM16.annotation.gtf")
> k <- keys(TxDb, keytype = "TXNAME")
> tx2gene <- select(TxDb, k, "GENEID", "TXNAME")
> head(tx2gene)

                TXNAME               GENEID
1 ENSMUST00000193812.1 ENSMUSG00000102693.1
2 ENSMUST00000082908.1 ENSMUSG00000064842.1
3 ENSMUST00000192857.1 ENSMUSG00000102851.1
4 ENSMUST00000161581.1 ENSMUSG00000089699.1
5 ENSMUST00000192183.1 ENSMUSG00000103147.1
6 ENSMUST00000193244.1 ENSMUSG00000102348.

I wasn't sure if this accomplished the same thing as indexing the GTF to a CSV and then reading that in, because of the error I get when running tximport: 

> sample_list <- read.delim("/Path/to/sample/list/20180214_fastq_ID_list copy.txt", sep = "\t", header = F)
> head(sample_list)
              V1
1   1_S18_Quants
2   2_S19_Quants
3   3_S20_Quants
4   4_S21_Quants

> files <- file.path("/Path/to/Quantsoutputfolder/SalmonQuants",sample_list[,"V1"], "quant.sf")

When I next tried to run tximport I got an error saying:

> txi <- tximport(files, type = "salmon", tx2gene = tx2gene)
reading in files with read_tsv
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 
Error in summarizeToGene(txi, tx2gene, ignoreTxVersion, countsFromAbundance) : 
  
  None of the transcripts in the quantification files are present
  in the first column of tx2gene. Check to see that you are using
  the same annotation for both.

So since I know I'm using the same annotation, I figure that somewhere along the way in making the tx2gene table I've done something wrong. I'm not sure the exact steps from the vignette and if I should be trying to use the ensembldb package. I were to use EnsemblDB would loading the most recent EnsDb.MMusculus.v79 accomplish the same goal after making a data frame including the Gene IDs? 

Thanks a lot, really appreciate the help!

Paul

___________________________________________________________________-

As an aside, when i first tried to run tximport I got the following error. I installed biocLite("rjson") and then ran the samples again. I only have biological replicates in my data files, so I am a little confused by the expectation that I have inferential replicates. 

reading in files with read_tsv
1 Error in readInfRepFish(x, type) : 
  importing inferential replicates for Salmon or Sailfish requires package `rjson`.
  to skip this step, set dropInfReps=TRUE

______________________________________

Also, as another aside, from my reading it sounds like Gencode and Ensembl are supposed to be the same annotation, but I tried to run Salmon on my data in parallel using the most recent Ensembl mouse transcriptome and got a much lower number of mappings that way. Anyone have an idea why that might be? 

The best way to start diagnosing this is to look at the names of the transcripts in the quant.sf files. Can you show some of these? You should then compare to the transcript names in tx2gene. These need to be the same. If there is an extra version digit(s) in the quant files, you should use the ignoreTxVersion argument.

Gencode and Ensembl use the same base set of transcripts but are definitely not the same. Just a small example is that Ensembl has identical transcripts on haplotype chromosomes, while Gencode does not include these duplicate transcripts (the duplicate txps cause problems for quantification actually).