Alternate Transcripts Paralogs

Question

[aj123]

Hello,

I have a few doubts related to how the best way is to go about finding these:

1. Finding unique regions when you compare various transcripts of a gene in one organism.
2. Finding evidences for a particular transcript.
3. Finding the % identity IN NCBI PROTEIN database for a particular sequence taken from uniprot. i.e i blast a protein sequence in uniprot database and get the paralogs for that along with percent identity. then i want to find the corresponding sequences for those paralogs in NCBI protein and find the %ID for my query sequence against those NCBI protein sequences. i tried ID mapping in uniprot and it gives me the corresponding IDs. but when i blast my query sequence against these IDs the %identity is way off from the %ID uniprot gives me.

any help/suggestions would be greatly appreciated.

[Damian Kao]

I am a bit confused. Do you mean you want to differentiate between alternative isoforms and paralogs?

[aj123]

No, i just want to find orthologs and paralogs for a particular gene from a particular species (human). Apart from this, I also want to find alternate transcripts for a gene in human.

Answer 1

[JC]

1. I assume that you want to identify isoform-specific exons for a gene, in that case, you need to overlap your annotations, and select the regions transcript-specific. You can do that with R:bioconductor and IRanges.
2. Evidence of what? expression? protein? conservation?
3. Are you using the same blast parameters? if the parameters are the same, the alignments must be similar.

[aj123]

@JC Thanks so much for your reply.

1) is there a way to do this using just online bioinfo databases like NCBI, Ensembl etc with out using IRanges? also, how would i a) find the annotations for a transcript b)go aobut overlapping my annotations c)select the regions which are transcript specific? i mean how would i know a region is transcript specific?

2) evidence of expression

3)yes, all blast parameters used are standard ones. im not sure still how/why this is happening

[JC]

1) and 2) depends on your model, human? mouse? 3) did you check if the alignments are the same? I'm thinking if you have 2 sequences, A with 100aa and B with 200aa, and you found a common alignment of 50aa, when you compare A->B the %ID is be 50% (based on A) but B->A is 25% (based on B)

[aj123]

Im using Human. Im not sure how you mean w.r.t the second part. im blasting always in a standard format. i.e original sequence vs. subject sequences. Still a little confused w.r.t finding the alternate transcripts for a gene.

[Jerven]

Default (standard) blast parameters differ between NCBI and UniProt blast sites. You have to be very careful about each setting.