How to get read-counts for sam files with transcriptome reference??

[Gazaldeep]

Hi,

I am working on rna-seq data analysis..
Mapped the reads to transcriptome reference (cDNA fasta file) for wheat using Bowtie2.
In the next step, I want to get the read counts.. I tried using featureCounts from Rsubread, but I get errors of the likes:

"
Warning: failed to find the gene identifier attribute in the 9th column of the provided GTF file.
The specified gene identifier attribute is 'transcript_id'
The attributes included in your GTF annotation are 'Parent=transcript:BAE47658;Name=BAE47658-1;constitutive=1;ensembl_end_phase=0;ensembl_phase=0;exon_id=BAE47658-1;rank=1;version=1'
"

But, the gff3 file i provided has trancript_id identifiers also.

In one case, I used GTF.attrType="protein_id" & GTF.featureType="CDS".. There was no warning in this case.. But the asigned reads were 0.

Also, featureCounts expects chromosome numbers in column 3 of sam file, but I have transcript ids as the reference was transcriptome..
What to use to get read-counts in this case??

Also, I wrote a perl script to count the number of times reads mapped to each of the transcripts (~15,000).. Can I use the counts obtained from this script?? Because I dont need to gather metafeatures into features and then count..

I am really confised, please help me out!

[GenoMax]

Also, I wrote a perl script to count the number of times reads mapped to each of the transcripts (~15,000).. Can I use the counts obtained from this script?? Because I dont need to gather metafeatures into features and then count..

That will work. How did you handle multi-mappers during alignment? That can skew some of your counts.

[Gazaldeep]

I used the --very-sensitive option in Bowtie2 while alignment. The use of this option was justified in (https://www.ncbi.nlm.nih.gov/pubmed/27735125 supplementary) to get read counts for homoeologs in wheat. So, this option leads to mapping a read against the best (single) match out of highly identical sequences also..
Is it alright or am I missing something?

Also, generally, when taking transcripts as reference, can't featurecounts be used to get read counts? What other softwares may be used? I'll continue using the in-house perl script for now.