We're working on generating proteomics datasets of non-model organisms which show considerable diversity between genera and for which there are several gene model sets available, of varying quality. Most proteome manuscripts use a single model organism and a canonical, well-annotated gene model set (e.g. from Uniprot) for peptide searches using MASCOT, SEQUEST, etc.
Is there any formal analysis available on the effects of using multiple combined gene model sets for peptide/protein matches, particularly in cases where the species of the sample and the species of the gene model set are not identical? In our case, I am considering combining several of the available gene model sets to form one large search database which I can use with samples that are different species (either from the same genus or related genera). Are there pitfalls in this approach, assuming one uses a protein-clustering algorithm to collapse redundant sequences? I would think that it may improve the number of proteins identified due to either a better match of the sample peptide and the gene model (due to a diverse set of models in the database) or by filling in gaps in individual gene model databases.
Hi Clinton
Personally, I am not aware of any formal analysis for the task. Currently, I am working on samples from fermentation on mixed bacterial species. For this purpose, I have combined the proteomes into one large FASTA as reference. Because they differ significantly, there are only a few overlapping proteins. I have not removed redundant sequences, as for my purpose, I prefer to get larger protein groups, where AC#s from the different organisms are not distinguishable. Then i link AC# with species in post-processing to get an idea about the redundancy and shared proteins between them.
But as you have several set for the same genus, it would b possible to remove redundant sequences. But if you don't, you can also get an idea of the quality of the model by back-tracking to each set. I see no problem with combining the sets - neither for the same genus or related genera. The better and more complete your reference, the better the output. However, you naturally should take precautions if the reference proteome becomes massive, as you increase the risk of false positives by chance. If you combine all the sets, how many AC#s would you have in the DB?
Also emerging are software tools that use several iterative database searching passes of the same data, combining multiple database search tools, searches with different post-translational modifications, and searches against different databases in an attempt to use each specific tool under ideal conditions, utilizing each for its specific strengths and integrating the results.
Ning K., Fermin D., Nesvizhskii A. I. (2010) Computational analysis of unassigned high-quality MS/MS spectra in proteomic data sets. Proteomics 10, 2712–2718 [PMC free article] [PubMed]
Ajit kumar Roy , you are absolutely right. I Know that the new version of Proteome Discoverer also includes this feature. Here you should be able to analyse your data using multiple search engine/scoring algorithms like MASCOT and SEQUEST in parallel and getting a combined output file. This shold be able to boost identification rates. The new version is landing in our lab next week, and we are very eager to test and correlate with e.g. MaxQuant which is my default analysis software.
- Badges
- Science topic
- Similar topics
- Linguistics
- Philology
- Classical Philology
- Manuscripts