I am trying to use tximport on my read counts from salmon to condense the ensemble transcript ID counts to gene ID counts.  This is what Ive tried:

library(GenomicFeatures)
txdb <- makeTxDbFromGFF("/path/gencode.v24.primary_assembly.annotation.gtf")
k <- keys(txdb, keytype = "GENEID")
df <- select(txdb, keys = k, keytype = "GENEID", columns = "TXNAME")
tx2gene <- df[, 2:1]
head(tx2gene)
             TXNAME             GENEID
1 ENST00000612152.4 ENSG00000000003.14
2 ENST00000373020.8 ENSG00000000003.14
3 ENST00000614008.4 ENSG00000000003.14
4 ENST00000496771.5 ENSG00000000003.14
5 ENST00000494424.1 ENSG00000000003.14
6 ENST00000373031.4  ENSG00000000005.5

The salmon quant.sf files dont have the decimal in the ensembl IDs so i just manually cut them off

a<- gsub("\\..*","",tx2gene[,1])
b<- gsub("\\..*","",tx2gene[,2])
c<-cbind(a,b)
colnames(c)=colnames(tx2gene)
tx2gene <- as.data.frame(c)
head(tx2gene)
           TXNAME          GENEID
1 ENST00000612152 ENSG00000000003
2 ENST00000373020 ENSG00000000003
3 ENST00000614008 ENSG00000000003
4 ENST00000496771 ENSG00000000003
5 ENST00000494424 ENSG00000000003
6 ENST00000373031 ENSG00000000005

library(tximport)
library(readr)
dir <- "/path_to_salmon_directory"
samples <- read.table("/path/file_names.txt", header=FALSE)
files <- file.path(dir,"salmon", samples$V1, "quant.sf")
names(files) <- paste0("sample", 1:9)
txi.salmon <- tximport(files, type = "salmon", tx2gene = tx2gene, reader = read_tsv)

reading in files
1 2 3 4 5 6 7 8 9 
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.

I have used all(file.exists(files)) to make sure the file paths are correct.  Also if I use read.table to import a single quant.sf file.  

quantfile=read.table(files[1])
head(quantfile)
               V1     V2              V3       V4       V5
1           dName Length EffectiveLength      TPM NumReads
2 ENST00000373020   2206         2063.46  31.9872  1495.67
3 ENST00000494424    820         677.463        0        0
4 ENST00000496771   1025         882.463 0.973862  19.4741
5 ENST00000612152   3796         3653.46  0.73197  60.5985
6 ENST00000614008    900         757.463  0.11619  1.99431

I can intersect the first column of the quant.sf file and the first column of the tx2gene file and I get nearly 200k matches.  I dont understand why the tximport function is saying there are no matches.  I have also tried using EnsDb.Hsapiens.v79 package to create the tx2gene file, and I get the same error. Any help appreciated!

It seems the problem is that you may have accidentally added a 'd' to the Salmon header. Because tximport supports import across different versions of Salmon, we can't assume to know which column is which without the names, and so it's important that the name of that column is actually "Name" and not "dName". Can you confirm this is a fix on your end?