Dear Jiahao,

Identifying the most relevant orthologs is a bit of an art.  I am providing Youtube links to Lectures 2 and 3 of NHGRI's Current Topics in Genome Analysis 2014 course, which should help you refine your search criteria.  

http://youtu.be/Z72nvSUtEng?list=PL1ay9ko4A8skSH3BBXU5LarkOn3nJzjn6

http://youtu.be/QE9TysFEQCA?list=PL1ay9ko4A8skSH3BBXU5LarkOn3nJzjn6

Best wishes,

Shaad

Dear Jiahao,

As mentioned above by Shaad, it is a bit of an art. We perform similar work, not in Schizophrenia, by determining potential homologs, orthologs...etc. from D. melanogaster to H. sapien, but we do this for hard to determine genes with small open reading frames (literally needles in many hay stacks aka pain the ass work)! Unfortunately it is not easy to the extent that you take a gene CDS and then blast against what you think should be the homolog or ortholog gene sequence or peptide sequence. This is obviously not for all genes but for now I will focus on an approach to find conservation between difficult to relate genes/ peptides.

When you break it down, you find that Drosophila peptides utilise different ratios of amino acids to that of Humans or most vertebrates. With this in mind, it will be very hard to blast pure Drosophila peptide sequence against Human sequences as you will find that structural motifs aren't entirely similar (at amino acid or nucleotide level), and domain positioning isn't always the same. One area that we have had luck in has been with the use of peptide structural alignment, using an online software called Phyre2. This has been mega helpful as it provides a good basis for 'homolog mining' :) but again you will need to do more to ensure similarity.

On top of this, we have had some luck with using consensus sequences whereby the Drosophila gene has been compared against other insects working your way up to vertebrates and higher vertebrates. You usually find that a consensus sequence that incorporates around 5-10 species on an evolutionary tree is the most likely hit to compare against human sequences (usually at the peptide level). But I must stress that this is a tricky area to get into as more often than most you find yourself producing a consensus product that is quite distant from the Drosophila original (or Human original) therefore artificial sequences may crop up. I don't personally do this as I don't have the expertise, but if you speak with a protein evolution professional you will see some sense in the madness.

So in closing this, I would suggest you utilise the hits you have but definitely compare this to a) structural similarity (i.e. protein structural motifs, folding potential...), b) consensus sequences of Drosophila based peptide sequences that incorporate 5-10 species between Drosophila and Humans (assuming conservation). I really hope this helps.

Best of luck,

Chris

Dear Jiahao,

The Drosophila RNAi Screening Center (DRSC) has developed an online tool that is designed to help get researchers started building a list of known or putative fly orthologs of human genes associated with diseases. The tool is called DIOPT-DIST and you can find it online at www.flyrnai.org/diopt-dist

At DIOPT-DIST, you can enter a fly gene (left-hand side) or you can enter a disease term or category (right-hand side). The results will include a list of fly genes, a list of human genes, and diseases, as well as information that helps you get a sense of the confidence level of the fly-human gene relationship (the DIOPT score can give some indication of this and if you click on that score in the results table, you can view protein alignments and see what protein domains are in common between the two proteins), and the confidence level of the human gene-disease relationship (such as whether the evidence is from the NCBI OMIM database or from GWAS data).

I hope this will be a helpful supplement to the excellent advice you have already received.

Best,

Stephanie

http://www.flyrnai.org/diopt-dist

Article An integrative approach to ortholog prediction for disease-f...

Dear Jiahao,

Just to add to the quite complete answers above I would also recommend the website Homophila (http://superfly.ucsd.edu/homophila/) as a port of call.

Best,

Tony

Dear All,

Thank you for all your responses! There are indeed very helpful:-)

Best,

Jiahao

ENSEMBL also has a great resource to get a quick list of ortholog genes of any two species for which genomic data is available. As mentioned by others, computational methods provide you only with putative orthologs. Unless the level of conservation between protein sequences of two species is very high, or someone else did the test, you´ll have to rescue mutant flies with the human ortholog.

Here are some great instructions how to get a conversion table from ENSEMBL:

https://www.biostars.org/p/15106/

With a bit of R programming or similar one can then generate a batch search if many genes need to be tested.

How about homologene?

https://www.ncbi.nlm.nih.gov/homologene/?term=pacs2