I'm looking for mRNA RefSeq sequences for a given gene in the UCSC genome browser by searching gene name "HLA-F" in human (scroll to Refseq genes section). There are several RefSeq mRNA for the gene, what is difference between these RefSeq sequences and which one of them is what I want?


I'll walk you through the process of determining this:

  1. Note that there are two gene names presented: HLA-F and HLA-F-AS1. The "AS" part means, "anti-sense", so it's on the opposite strand...so ignore all of the HLA-F-AS1s.
  2. Note the associated chromosomes. chr6 is a normal chromosome, but things like chr6_cox_hap2 aren't. Instead, these are haplotype patches/chromosomes. The HLA regions are highly variable, so having only a single reference sequence ends up being problematic in many analyses, such as finding sequence variants. For your purposes you may or may not want to ignore these, it depends on what your goal is.
  3. If you ignore the haplotype patches then you're left with 3 listings on chr6. These are the 3 isoforms of HLA-F. You end up needing to use the refseq IDs for each of these.

BTW, Gencode/Ensembl list 4 isoforms for HLA-F, rather than 3. Whenever UCSC disagrees with Gencode or Ensembl, go with Gencode or Ensembl.

  • $\begingroup$ Many thanks for your nice reply. Just one thing, would you please help me out how to get Refseq sequences of interest from Ensemble in a way that the CDS is recognizable from UTRs in the sequences? $\endgroup$ – Mary Sep 4 '15 at 21:19
  • $\begingroup$ Google "Biomart". $\endgroup$ – Devon Ryan Sep 4 '15 at 21:54
  • $\begingroup$ Would you mind justifying your last statement? >Whenever UCSC disagrees with Gencode or Ensembl, go with Gencode or Ensembl. Not that I disagree, but it would be better to give a reason for suggesting some source is better than another. $\endgroup$ – ddiez Sep 6 '15 at 14:40
  • 2
    $\begingroup$ UCSC's annotations have historically been a bit of a mess. For example, they have genes with the same ID on multiple chromosomes and strands, which will break a number of tools. refSeq in particular tends to be overly conservative in what it includes, so typically using it results in missing a number of likely correct isoforms. For what it's worth, UCSC now defaults to Gencode for human (and Gencode is essentially the same as Ensemble) $\endgroup$ – Devon Ryan Sep 6 '15 at 20:51
  • $\begingroup$ @DevonRyan Thanks. Maybe this paper is also relevant. $\endgroup$ – ddiez Sep 7 '15 at 2:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.