I am interested in detection of all protein isoforms arising from alternative splicing events. To do this, I would like to set up database searches to account for all isoforms by parsing unique peptides separately from razor peptides common to 2 or more isoforms. In the end it would result in separate protein assignments for unique isoforms and for common isoforms. Does anyone do this already, and if so, is there a way to set database search parameters in such a way as to do this quantitatively, i.e. unique isoform 1 protein ratio, unique isoform 2 protein ratio, protein ratio arising from common peptide isoforms 1 and 2, etc.? Even standard parsimony assumptions do not always result in what I consider accurate protein ratios. For instance, looking at a protein where isoform 2 is subsumed by isoform 1 I've detected isoform 1 unique peptides at one ratio and common 1/2 peptides at a very different ratio. By standard parsimony I would conclude that only 1 is present, but by the peptide ratios it is clear that 2 is also present but at a very different ratio than isoform 1. I've read several papers touching on these issues, but haven't yet found anything that fully addresses them.
Identifying and quantifying multiple protein isoforms is best accomplished by setting up an MRM approach, where you use heavy standards of each isoform to establish your assay and then measure your samples.
For the reasons you mention in your question, I do not think that a non-targeted approach will give you the results you seek.
What other information you know about your gene? Gene structure? How many exons? size of gene? type of AS, exon skipping, intron sliding? Do some predictive analysis? AS may be condition specific...or dormant...or environment specific...sample plays a role here...species data is key...rate of AS in the species organism...
Thanks for the suggestions, Sonja and Pandjassarame. The initial experiment from which I detected AS events includes 2780 proteins identified at an estimated FDR of 0.2%. The next experiment has >5000 IDs, so I am hoping to automate the search for AS variants to avoid sifting through one by one. Unfortunately, Sonja, I don't have specific target proteins, so setting up MRM is not an option. As for the various conditions you describe Pandjassarame, I am relying on whatever information is available in the UniProt database to define different protein variants. Fairly well curated, but undoubtedly could include additional information. As mentioned, I don't have specific targets, but rather I am hoping to detect any AS events from a discovery proteomics standpoint. Because of the experiment I am doing, I know there will likely be several which could be critical to understanding the question I am trying to answer.
I assume you are dealing with human data...if so divide your data set based on number of exons in each gene as available with the CDS of genome data...now you will start from exon = 1; to 2, to 3, to others...so you will have different sets having different gene structure of varying number of exons as given in CDS of genome sequence by gene prediction programs...now you deal with different sets separately there after...What type of tissue samples you are dealing with? why? how? ask these...you will reach the next step....grouping, sub-grouping based on gene structure followed by alignment will help to converge to some extend....further you change tissue or condition...repeat...compare...your variants may fairly change ...your problem is to do with gene structure and in this case Uniprot could hardly help...
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Molecular Biological Techniques