Hello,
I am wondering what is the best method for searching very remote homologs of human proteins. For example, if I am interested in protein XY, I would download all protein sequences of XY from NCBI (limited to Eukaryote domain), reconstruct ancestral protein sequence (for example in Phylobot (http://www.phylobot.com/)) and do protein blast or HMMER against database of all known "hypothetical" protein sequences?
Is it sufficiently sensitive approach? Or does somebody know better variant? I am interested in searching remote eukaryotic protein homologs in domain Archea.
Last but not least I would like to ask, if I eventually find such "putative homologous protein" with about 10 - 15 % amino acid residues identity, is there any bioinformatic way for validating such result? For instance, look at conservation level of these amino acid which are suggested to have important function in Eukaryota (cysteins, arginines, etc.)? Is there any automatic workflow or user-friedly tool to do this?
Thank you for all answers
Hi Martin,
Basically yes.
1) With the sequence or identifier of your XY protein run directly PSI-BLAST.
( https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE=Proteins&PROGRAM=blastp&RUN_PSIBLAST=on .)
You can try running it with the following options:
a) matrix: BLOSUM45 or with BLOSUM62 that comes by default but the lowest the score more distantly related proteins you may get. For more info look at :
https://en.wikipedia.org/wiki/BLOSUM#Score_of_the_BLOSUM_matrices
b) It may be a good idea to run it filtering out the low-complexity regions
c) Change the Max target sequences reported depending or what you are interested on.
d) Database: you can use the non redundant proteins as default or to see if any of the matches has a PDB solved structure, select PDB (I will run it first using PDB, as if there is any match with a PDB structure you most surely will like to do a 3D visualisation…).
2. Then, after the first run of psi-blast, select the new matches to include and run it again to find homologs to these new sequence. Keep running psi-blast iteratively on each run select the new hits you want to include in the search until you consider you should stop based on the number of hits, the query cover (
If the homologs are too remote then it's not possible to detect them based on their protein sequence. However, the 3D structure will still show overall similarities. I don't know if there are good databases to search based on the 3D structure but for many protein families there are no structures solved anyway, so this won't be applicable.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biota
- Eukaryota