reference mapping database in RNA-seq, which one is best? UCSC or Ensembl?

[super0925]

Hi everyone, I am a rookie in RNA-Seq.
When I do the reference mapping in RNA-seq, what is the advantage of each reference genome database(UCSC, NCBI or Ensembl)?
Which one could I use? or no matter what I use?
Do I need to uniform the database in different species (e.g. all species use the UCSC database)?

http://cufflinks.cbcb.umd.edu/igenomes.html

[GenoMax]

I don't think there is a "best" database. At a minimum you should plan to stick with one source for all analysis.

UCSC's annotations are conservative as compared to the Ensembl (which may have several computational predictions in the genome annotation GTF files).

Illumina has an expanded iGenomes source available here: http://support.illumina.com/sequenci...e/igenome.ilmn

[super0925]

I don't think there is a "best" database. At a minimum you should plan to stick with one source for all analysis.

UCSC's annotations are conservative as compared to the Ensembl (which may have several computational predictions in the genome annotation GTF files).

Illumina has an expanded iGenomes source available here: http://support.illumina.com/sequenci...e/igenome.ilmn

Thank you for your remind.
I use Btau_4.0 (Ensembl) in cow but hg19 (UCSC) in human.
It seems that I need to make cow and human both stick to Ensembl (or UCSC).

[GenoMax]

It seems that I need to make cow and human both stick to Ensembl (or UCSC).

Not strictly. They are two different genomes, after all.

I just was pointing out that there are annotation differences between UCSC/Ensembl. If you are planning to use tuxedo suite you may run into some problems with the GTF files because of the content differences.

[super0925]

Not strictly. They are two different genomes, after all.

I just was pointing out that there are annotation differences between UCSC/Ensembl. If you are planning to use tuxedo suite you may run into some problems with the GTF files because of the content differences.

Sorry I am confused. So do you suggest me all (human, cow and mouse) reference genome to use UCSC if I use Tuxedo? or all dataset to use NCBI ref? Or all to use Ensembl ref? Or actually it doesn't matter but I must stick to one database among different species.
Thank you!

[GenoMax]

If you intend to try to map results across various species then it may be useful to use UCSC since they will have "liftOver" files available for cross species mapping of regions.

Otherwise as long as your species specific analyses are independent you can select and stick with one (any of the three) source of data.

[super0925]

If you intend to try to map results across various species then it may be useful to use UCSC since they will have "liftOver" files available for cross species mapping of regions.

Otherwise as long as your species specific analyses are independent you can select and stick with one (any of the three) source of data.

Thank you！
Do You mean
condition 1 vs condition 2 in cow
condition 1 vs condition 2 in human
I could use any of 3 source of data.

But if condition 1 in cow vs condition 1 in human
UCSC is best.
Am I right?

[GenoMax]

Thank you！
Do You mean
condition 1 vs condition 2 in cow
condition 1 vs condition 2 in human
I could use any of 3 source of data.

But if condition 1 in cow vs condition 1 in human
UCSC is best.
Am I right?

Yes.

For second case, UCSC is "best" only in the sense that it will allow you to possibly map regions across the species easily using liftOver data.

[super0925]

Yes.

For second case, UCSC is "best" only in the sense that it will allow you to possibly map regions across the species easily using liftOver data.

I found a online tools to convert Ensembl ID to gene symbol
Biomart
Is this good to solve the limited in Ensembl ?
And I found that some genes with Ensembl ID but without gene symbol.

[GenoMax]

I found a online tools to convert Ensembl ID to gene symbol
Biomart
Is this good to solve the limited in Ensembl ?
And I found that some genes with Ensembl ID but without gene symbol.

Ensembl data tends to have computational/automated annotations that may not be fully validated. As a result they will not have gene symbols (and/or there may be multiple entries for a region).

[DerSeb]

For mapping I always use the Ensembl GTF annotation files, since they are less conservative and contain many possible transcripts.

I simply want to maximize the amount of splice junctions available for mapping.

For later analyses I decide depending on project/organism which annotation files I use.

[super0925]

Ensembl data tends to have computational/automated annotations that may not be fully validated. As a result they will not have gene symbols (and/or there may be multiple entries for a region).

Hi if I have a list of genes by Ensembl ID or Gene namein cows, I want to covert them to human Ensembl ID and Gene name. I could do it by website http://www.ensembl.org/biomart/martview/ AM I right?
But it is not automatically.
Could you please recommend other R package to achieve the Gene-ID converter?