Good day, I am doing analysis of my RNA-Seq data using Salmon for quantification and STAR for alignment. Both tools require building a reference index. The GRCr8 rat reference genome files are from ensembl: https://www.ensembl.org/Rattus_norvegicus/Info/Index

For Salmon, I followed the advice of building a decoy-aware index by combining the cdna library (https://ftp.ensembl.org/pub/release-115/fasta/rattus_norvegicus/cdna/#:~:text=Rattus_norvegicus.GRCr8.cdna.all.fa.gz) with the genome primary assembly. My problem is with the genome primary assembly file -- the available GRCr8 primary assembly files are separated per chromosome (https://ftp.ensembl.org/pub/release-115/fasta/rattus_norvegicus/dna/).

Hence, I concatenated the per-chromosome primary assembly files (non-chromosomal, chromosome 1-20, MT, X, Y) into a single primary assembly file by simple cat command. Then I built the decoy-aware index following the instructions here: https://combine-lab.github.io/alevin-tutorial/2019/selective-alignment/

I also built my STAR index using my concatenated primary assembly file.

However, there is a huge discrepancy in the aligned reads: only ~33% in Salmon and ~90% in STAR.

I'd like to ask if anyone has experience building an index from the GRCr8 rat reference genome and what are the expected results for aligned reads in Salmon and STAR. Would the alternative options be the best:

Salmon:

1. Use only the cdna library as the index (not decoy-aware)
2. Use the toplevel file for the genome reference in building a decoy-aware index, instead of concatenating the per-chromosome primary assembly files.

STAR:

1. Use the toplevel file for the genome reference, instead of concatenating the per-chromosome primary assembly files.

Thanks in advance