Hello,

I have a dataset contains the following columns: transcript_id, gene_id, length, effective_length, expected_count, TPM, FPKM, and IsoPct. And for the same gene_id more than two transript_id .. so it's a transcript-level dataset

I thought I need to use tximport to convert it to counts so I could use it for DESeq2, but I didn't really get how can I convert it exactly.

The output file you have appears to be generated by RSEM. In that case, you will have a couple of different files per sample available - sampleID.genes.results and sampleID.isoforms.results. Indeed, tximport can do the importing and bias adjustment for you. See the RSEM section of the vignette for an explanation of how to do it.