RNA-seq mapping Reference genome

Question

[k.kathirvel93]

Hi EveryOne,

I am using STAR and HISAT2 and two more tools RNA-seq data analysis. My question is, should I align my Draft genome with GRCh38 (.gtf) reference annotation or genome (.fa)reference? Which one is best?

[Arup Ghosh]

Draft genome sequence and transcript level? Can you elaborate more about the situation? If you are talking about alignment in general, the gene transfer format does not include any sequence-related information, so for a draft genome sequence alignment, the ref genome is the right file. The GTF file is used by some tools as a guide to restrict the analysis to known transcripts only (reduces the time required for analysis).

[Jeffin Rockey]

Could you elaborate on the draft genome and the other genome (.fa)reference? Usually there will be only one genome, one gtf and one or more fastq files.

[k.kathirvel93]

I am doing Transcriptome analysis. Draft genome - sequenced from patient sample. genome(.fa) - GRCh38 genome reference fasta file. Very clearly, STAR is taking .gtf and genome.fa file for reference mapping. But HISAT2 is taking only genome.fa as reference. Which one is correct. Thanks

[ATpoint]

That is not true. Hisat2 needs a file of known splice junctions, which you have to generate from a reference GTF file. You could run Hisat2 without that file, but then it loses its splice-awareness, which is essential for meaningful alignment of RNA-seq reads that span exon-exon junctions. You can use both tools. My recommendation for you is the following: Look into the usage of both tools and choose the one that you feel more comfortable with. Both tools are well-accepted, tested and produce meaningful results. Much more important than the alignment is the downstream analysis, which you should focus on. What exactly is your final goal?

[k.kathirvel93]

My aim is differential gene expression analysis. I want to use both of the tools for comparison. Can i get the code for HIsat2 indexing with both genome(.fa) and annotation(.gtf) reference? Thanks.

[ATpoint]

./hisat2-build will give you the information on the indexing.

./hisat2_extract_splice_sites.py extracts the splice sites from the GTF.

Answer 1

[Devon Ryan]

I'm not sure why you bothered making a draft genome of a patient, but you'd be better off not using it when analysing your RNAseq data. Just align to the reference genome and use one of the reference GTF files.