[DonMauricio]

Sorry but I dont undertand
¿are the rDNAs in both databases too short?
average size is arround 110nt, but 28S has 5070nt and 18S has 1819 nt
¿¿??

[apredeus]

for mm10, there are no more rRNA elements in random chromosomes or scaffolds - they have all been placed in regular chromosomes.

mm10_rRNA_rmsk.gtf has 1564 intervals covering 168298 bp
mm10_rRNA_gencode.gtf has 353 intervals covering 39472 bp (vM2 of mouse Gencode)

They overlap on 33.2 kbp, and the only intervals in Gencode that are not present in rmsk at all follow:

chr2	61798180	61798296	Gm25966
chr6	94826781	94826923	Gm24601
chr14	12427532	12427641	n-R5s45
chr18	73533403	73533550	n-R5-8s1

Again, one can merge rmsk with these intervals and get the most complete coverage.

[apredeus]

so, from the comparison I've done, it seems like rmsk intervals have a lot more coverage. They include virtually all of the Gencode intervals, but many more unique intervals as well.

So, for hg19:

hg19_rRNA_rmsk.gtf has 1769 intervals covering 193760 bp
hg19_rRNA_gencode.gtf has 571 intervals covering 70960 bp (v19 of human Gencode)

When you intersect the two, you see that they have about 50 kb in common. However all of the gencode intervals but the few that follow are included in rmsk version (I can't evaluate differences in random chromosomes and unplaced scaffolds since they have different names):

chr2	133010727	133010878	RNA5-8SP5
chr2	162266065	162266181	5S_rRNA
chr9	32293556	32293690	RNA5SP281
chr9	110681147	110681259	RNA5SP293
chr9	111754689	111754849	RNA5-8SP3
chr10	49248476	49248591	RNA5SP315
chr11	8866810	8866905	RNA5SP330
chr11	96207736	96207856	RNA5SP346
chr12	66460001	66460118	RNA5SP362
chr15	98015356	98015482	RNA5SP401
chr16	33965426	33965577	RNA5-8SP2
chr19	24187160	24187309	RNA5-8SP4
chr20	5326652	5326806	RNA5-8SP7
chrY	10037764	10037915	RNA5-8SP6

So basically, I'm going to use rmsk gtf file with the addition of these 14 lines. This should be more than enough for my purposes. I'm also definitely including all of the random chromosomes etc. since a lot of rRNA elements are there.

[apredeus]

In terms of the Ensembl vs Hg19 (rmsk) coordinates/annotations for rRNA, although I am new to this, it doesn't seem like they should necessarily correspond. Don't they have different annotation pipelines/procedures, anyway?

yes, they are different, but the question is which one is more comprehensive (and thus will be more complete in i.e. evaluation of rRNA presence in RNA-seq experiment).

Actually, as I'm finding out, they are super different. I'm not quite sure what is the reason for that. I'll post more details in the next post.

[jbchang]

Thanks, guys, for this valuable discussion. After looking through the rmsk rRNA entries more closely, they seem to make enough sense that I'm willing to trust RepeatMasker.

In terms of the Ensembl vs Hg19 (rmsk) coordinates/annotations for rRNA, although I am new to this, it doesn't seem like they should necessarily correspond. Don't they have different annotation pipelines/procedures, anyway?


Best,
Jeremy

[GenoMax]

Right, yes, I've considered this. From what I've learnt, GRCh37 should be precisely identical in terms of genomic coordinates to what's known as hg19 in UCSC notation.

Does the same hold true for GRCm38/mm10?

That is correct.

[apredeus]

Right, yes, I've considered this. From what I've learnt, GRCh37 should be precisely identical in terms of genomic coordinates to what's known as hg19 in UCSC notation.

Does the same hold true for GRCm38/mm10?

[GenoMax]

Now, I'm seeing quite a bit of differences between intervals provided in Ensembl GTF file and in UCSC tables. I'll have to look at this in more detail to try and understand where do they come from.

First thing to check would be the genome builds and make sure they are the same.

[apredeus]

So, to continue, I've looked at the "gene_id" identifiers provided in UCSC "rmsk" table (which, as I understand now, refers to RepeatMasker database, go me! ).

For hg19, the output is

   1275 5S
    414 LSU-rRNA_Hsa
     80 SSU-rRNA_Hsa

For mm9, it is

      1035 5S
    491 LSU-rRNA_Hsa
     46 SSU-rRNA_Hsa

So basically the conclusions is 1) UCSC DOES include all of the rRNA into their "rmsk" table annotation, and 2) to get realistic picture of rRNA presence, you should have genome with all of the "random" chromsomes, etc.

Now, I'm seeing quite a bit of differences between intervals provided in Ensembl GTF file and in UCSC tables. I'll have to look at this in more detail to try and understand where do they come from.

[apredeus]

Hey folks,
I've been looking into this for a day or two as well, and I also downloaded the .gtf corresponding to repClass=rRNA (also tRNA) from the UCSC Table Browser.
When I scroll through the .gtf file, though, it seems to only have entries corresponding to the 5S ribosome. Is this true for you guys, too?
Thanks,
Jeremy

So I spent some time to learn the situation, and this is actually pretty cool
So, there are 6 kinds of ribosomal RNA in mammals (well for sure in humans): 3 belonging to large subunit (LSU rRNA: 5S, 5.8S, and 28S), 1 belonging to small subunit (SSU rRNA: 18 S), and 2 mitochondrial rRNAs (12S, 16S).

Mitochondrial ones are the easiest - they reside, well, in chromosome M Two rRNAs, two genes, very neat.

However, others are a real mess. Here's what Wiki says about it:

The 28S, 5.8S, and 18S rRNAs are encoded by a single transcription unit (45S) separated by 2 internally transcribed spacers. The 45S rDNA is organized into 5 clusters (each has 30-40 repeats) on chromosomes 13, 14, 15, 21, and 22. These are transcribed by RNA polymerase I. 5S occurs in tandem arrays (~200-300 true 5S genes and many dispersed pseudogenes), the largest one on the chromosome 1q41-42. 5S rRNA is transcribed by RNA polymerase III.

So, yes, we do see a ton of 5S genes and pseudogenes all over the annotated GTFs. There are also some 5.8S, but not many (about ten). Where are the the above-mentioned clusters of 45 rDNA though? Well, that's when it gets interesting. They are still not annotated! Gene cards say the following here:

The sequences coding for ribosomal RNAs are present as rDNA repeating units, designated RNR1 through RNR5, in the p12 region of chromosomes 13, 14, 15, 21 and 22. A 45S rRNA which serves as the precursor for the 18S, 5.8S and 28S rRNA, is transcribed from each rDNA unit by RNA polymerase I. The number of rDNA repeating units varied between individuals and from chromosome to chromosome, although usually 30 to 40 repeats are found on each chromosome. These ribosomal repeating units are not currently annotated on the reference genome. This gene represents the portion of one rDNA repeat which encodes an 18S rRNA.(provided by RefSeq, Mar 2009) .

Indeed, in Ensembl file I've found the following entries:

GL000220.1      109078  110946  5S_rRNA rRNA
GL000220.1      109078  110946  RNA18S5 rRNA
GL000220.1      109078  110946  RNA18S5 rRNA
GL000220.1      112025  112177  RNA18S5 rRNA
GL000220.1      112025  112177  RNA5-8S5        rRNA
GL000220.1      112025  112177  RNA5-8S5        rRNA
GL000220.1      113348  118417  RNA5-8S5        rRNA
GL000220.1      113348  118417  RNA28S5 rRNA
GL000220.1      113348  118417  RNA28S5 rRNA
GL000220.1      114151  114242  RNA28S5 rRNA
GL000220.1      114151  114242  RNA28S5 rRNA
GL000220.1      118197  118253  RNA28S5 rRNA
GL000220.1      118197  118253  RNA28S5 rRNA
GL000220.1      155997  156149  RNA28S5 rRNA
GL000220.1      155997  156149  RNA5-8S5        rRNA
GL000220.1      155997  156149  RNA5-8S5        rRNA
GL000228.1      20113   20230   RNA5-8S5        rRNA
GL000228.1      20113   20230   5S_rRNA rRNA
GL000228.1      20113   20230   5S_rRNA rRNA
GL000228.1      22673   22791   5S_rRNA rRNA
GL000228.1      22673   22791   5S_rRNA rRNA
GL000228.1      22673   22791   5S_rRNA rRNA

As of GRCh38, contig GL000220.1 is still unplaced. However, GL000228.1 is obsolete and probably is present in main GRCh38 assembly.

Well, I think we all learned something today!

[jbchang]

Hey folks,

I've been looking into this for a day or two as well, and I also downloaded the .gtf corresponding to repClass=rRNA (also tRNA) from the UCSC Table Browser.

When I scroll through the .gtf file, though, it seems to only have entries corresponding to the 5S ribosome. Is this true for you guys, too?


Thanks,
Jeremy

[apredeus]

There are differences in GTF files for UCSC and Ensembl downloaded from the iGenomes. See for reference: http://seqanswers.com/forums/showthread.php?t=41701

Can you check to see if the data you got from UCSC matches the Ensembl GTF?

I definitely will, and I'll post the results here in a day or two.

[GenoMax]

There are differences in GTF files for UCSC and Ensembl downloaded from the iGenomes. See for reference: http://seqanswers.com/forums/showthread.php?t=41701

Can you check to see if the data you got from UCSC matches the Ensembl GTF?

[apredeus]

Cool. It would be interesting to compare the two.

[GenoMax]

As another option: this information is also in the GTF files available from Ensembl: http://www.ensembl.org/info/data/ftp/index.html You will have to "grep" out the information.