I'm analysing RNA-Seq data of cod (Gadus morhua) which is a non-model species, meaning that the quality of its genome and annotation is not as good as model species such as human and mouse. The downstream analysis of the RNA-Seq alignment is to do differential expression analysis and pathway enrichment analysis. I see that for model species, the mainstream is to align to genome (a little bit confused here, since an answer under this post said that the mainstream of model genome is to align to transcriptome, but I found the opposite based on some papers and published RNA-Seq workflows). But how about the non-model species?
When I aligned the RAN-Seq reads using HISAT2, I tried aligning to both genome (download DNA fasta file and annotation gtf file from ENSEMBL) and transcriptome (download cDNA fasta file from ENSEMBL). I found that the alignment rate to genome was over 90% but the one to transcriptome was only around 40%. Does that mean that it's better to use genome instead of transcriptome? I also tried using Qualimap to analyse the BAM files generated from two alignments, but I didn't see much difference.