Question with Salmon quant and tximport

I am fairly new to RNA-seq, I am trying to compare transcriptomes across various samples. I am having difficulty with the use of tximport to merge my salmon files.

May I know is it ok to merge the TPM column from each salmon file manually, to generate a combined file for DGE analysis like Degust (my salmon files already have transcript ID)?

I am worried because the samples actually have different sequencing depth (some timepoint have double the sequencing depth), do I need the tximport “scale up to library” function to actually make sense of my RNA-seq results, or will later DGE analysis in Degust like edgeR or voom took care of the normalization problem?

I am having difficulty in the tximport because I am doing dual-RNA seq, and I am aligning my libraries to a combined reference transcriptome of two organism, after the alignment, the gtf annotation file that was also a combined gtf file was not compatible somehow.

Any help would be appreciated. Many thanks!