How to get TCGA transcript isoforms sequences ?

[watermark]

Hi all,

I've downloaded the breast cancer RNA-seq2 data from TCGA. it's contain the isoform of the genes .
I'm looking way to get the sequence of theses isoforms. I tried BioMart and UCSC ,but I couldn't fin it. I face with error like: no such ID exist.

Can anybody help me ?
my ids are like :

isoform_id
uc011lsn.1
uc010uoa.1
uc002bgz.2
uc002bic.2
uc010zzl.1
uc001jiu.2
uc010qhg.1

[GenoMax]

http://www.biostars.org/p/93395/

I see now that you want the sequences not just ID's.

[watermark]

Yes, exactly, I want to have the sequence of those IDs, which is the transcript isoforms from TCGA.

[GenoMax]

Here is a BED format file for the transcripts used for TCGA analysis: https://webshare.bioinf.unc.edu/publ...A/unc_hg19.bed

General description of the TCGA pipeline: https://webshare.bioinf.unc.edu/publ...eq_summary.pdf

[dpryan]

Ah, I guess you're the "Jack" over on Biostars that PM'd me. There may be an easier way to do this, but one method would use the BED file that Genomax posted. That file actually have the coordinates of the exons for each of the transcripts, though that may not have been obvious at first glance. From that, one could create a GRanges object in R (see the GenomicRanges package) and then use getSeq() to just get the sequence for each exon (getSeq() is from BSgenomes, so you're using a good number of different packages). You could then "just" apply a function to output the merged sequence (this is only easy if you're really familiar with R).

The hardest part of this is likely parsing the BED file to get all of the exonic coordinates. Let's just take a single example:

chr15   82647285        82707815        uc002bgz.2      0       +       0       0       0       5       307,57,140,80,1010,     0,45585,49532,53207,59520,

The last column gives 5' offsets inside the coordinates specified in the first 3 columns for each exon. The next to last column gives the width of each of those exons in bases. In other words, this transcript has 5 exons starting at position:

#1: 82647285+0 = 82647285
#2: 82647285+45585= 82692870
#3: 82647285+49532= 82696817
#4: 82647285+53207= 82700492
#5: 82647285+59520= 82706805

The easiest way to derive that is probably to split that next to last column by commas, drop the last value, convert to numeric, c(start_position, that_vector), and then use cumsum(). The output would then be the appropriate positions. You don't need to specify the end positions of each exon, just the widths, so that part is much easier.

It'd be good to double check that that works with a transcript on the - strand, since sometimes UCSC does weird things with the coordinates there.

[GenoMax]

@Watermark:

Sequence files for the isoforms are available here: https://webshare.bioinf.unc.edu/publ...transcripts.fa

Mapping file: https://webshare.bioinf.unc.edu/publ...ownToLocus.txt

[watermark]

Thanks GenoMax.

[watermark]

Thanks dpryan. it was helpful.

[watermark]

@Watermark:

Sequence files for the isoforms are available here: https://webshare.bioinf.unc.edu/publ...transcripts.fa

Mapping file: https://webshare.bioinf.unc.edu/publ...ownToLocus.txt

Do you know how can I find which miRNAs binds to theses isoforms ?