How to get human cDNA sequences together with UTR regions?

Question

[Apex92]

Dear all,

I have downloaded the human genome and gtf files from Gencode. Based on these two files I want to generate a fasta file that has cDNA sequences including 5' and 3' UTRs for protein-coding genes only. What is the simplest and fastest way to do this?

Would this script work? gffread -w transcripts.fa -g genome.fa transcripts.gtf

Thank you.

[i.sudbery]

cDNA is the "complementary" DNA to the mRNA transcript. mRNA transcripts include UTRs, so the cDNA sequence should too.

Answer 1

[Istvan Albert]

The command you specify will concatenate the exon sequence for each transcript and as long as the exons contain the UTRs you will get those.

To keep only protein-coding exons, you might need to preprocess the GTF file to keep only those that have gene_type "protein_coding" tag.

In general, though I would recommend downloading the CDNA files from the same source and filtering that with some other method.

[i.sudbery]

This probably wants to be transcript_biotype, not gene_biotype as its possible to have non-coding transcripts of coding genes.

Answer 2

[Apex92]

Thank you for your comments. At the end of the day, I tried this approach so this might be helpful for others as well. And I would also be happy to get your feedback in case I encountered any mistakes.

1. I downloaded both genome and gtf files from Gencode.

2. Preprocessed the gtf file and converted it to a bed format in the structure below (keeping only protein-coding transcripts): chr start end transcript_name type strand.

3. Used bedtools to extract sequences in fasta format from the genome file using bedtools as
   bedtools getfasta -fi genome.fa -bed gencode_protein_coding.bed -name > hsa_protein_coding_transcripts.fa

[Istvan Albert]

as far as I know the bedtools getfasta can only concatenate exons if you had it in the 12 column format with block information,

if all you had was the 6 column BED as you describe it, then how could it identify the exons that form a transcript?

I believe the method that you describe will generate the unspliced transcript