I have 9 amino acid sequences from experimental gene expression of which we have no 3D structural data. The sequences do not exist in any of the online databases. Running PBlast we only see similarity from a few known proteins (1, 13, 11, 34, 52, 31, 3, 76 and 47, respectively for each of the nine sequences), with an "OK" score but poor coverage e.g., 64% max, median of around 32% coverage. This doesn't allow enough data for evolutionary contact prediction such as Gremlin (I've tried, no results yielded), and the only route I currently know of at this point is to use one of the model-free/template-free prediction tools (CASP11) e.g., I-TASSER. From doing so, only one of the sequences yielded an "OK" (just my judgement) C-Score.
Is there anything/any tools I can do/use which takes what little PBlast info there is and yields a 3D structure of the protein, yet not any of the CASP11 template-free tools?
Thanks
I forgot to mention, as these are full protein sequences most of them are in the region of 200 to 250 amino acids long, thus, from the documentation, PEP-fold won't work.
You may try using the QUARK server for your sequences which are of ~200 amino acids length to refine the probable poorly predicted regions of your 3D models generated from I-TASSER (you may use the secondary structure prediction information for your sequences as a reference)
http://zhanglab.ccmb.med.umich.edu/QUARK/
Hi Philippe,
I had a good read of your recent publication. Sadly, after attempting the first stage of your pipe-line it turns out that Pfam yields no results for my 105 amino acid sequences.
Again, a real shame, as it looks like I may be restricted to pure de-novo modelling unless the experimentalists move from doing micro-array based work to protein structure (I very much doubt they will - they are genomics and bio-informaticians).
Thanks again
What database you used to find homologs? NCBI and UniProt have mostly proteins from genomic projects, very few from metagenomics ones. JGI database has almost 400x (predicted) proteins, so for some proteins you could find many more homologs there. If you find same, then you could run PSI Blast to search for more distant homologs
- Badges
- Science topic