My PI has assigned me a task to gather evidence to determine whether or not a list of particular species of Campylobacter and Helicobacter code for a list of several proteins. To put it simply I’ve been asked, “Do species X, Y, Z code for proteins A, B, C?” The genomes of these bacterial species have been fully sequenced and I am able to retrieve the FASTA files for each of them. I am having trouble deciding on the proper method to screen these genomes for the genes. I have been given the names of the proteins, i.e. VgrG and HCP, but the amino acids sequences listed on databases are annotated to different species of the bacteria. I seem to be stuck in a grey area and am having trouble connecting the overwhelming amount of resources available on the NCBI website. I have not found a specific guide to perform this task - Could you please provide me with a solution to my problems?


First of all, don't reinvent the wheel, search for annotated homologs first. Assuming you don't find them, the next step is:

  1. Collect the sequences of your query proteins (not genes) from one species in a multifasta file.

  2. Run a tBLASTn with those sequences against the genomes of all the other species of interest.

  3. Analyze. Look for HSPs with a certain level of sequence identity. The details of what threshold you choose will depend on your data, I can't help you there.

While the tBLASTn approach is going to be more sensitive, an easier method that can find gene structures and not just HSPs is to use exonerate:

exonerate -m p2g proteins.fa genome.fa

Finally, for a more sensitive search than either of the above, useful for finding more distant homologs, run a Psi-tBLASTn of your proteins against the genomes of interest.

In all of these approaches, the tricky bit is to choose the right threshold. Unfortunately, there is no golden rule here. It all depends on how closely related your species are and how similar the homologs. You will have to try a few different ones before you can choose the right one.

  • $\begingroup$ I am unclear as to how I could go about obtaining these query proteins. They are not listed in any databases yet. I think my PI wants me to predict whether or not a protein exists in these organisms. $\endgroup$ – codax Oct 4 '15 at 22:27
  • $\begingroup$ @codax that's a completely different question! What are you trying to do? WHat do you have and what are you looking for? Don't you have a list of proteins in species X and want to find them in species Y? $\endgroup$ – terdon Oct 4 '15 at 22:34
  • $\begingroup$ I'm really sorry if I was unclear, but your advice is invaluable to me and I'm learning so much. I have a list of species and I need to find if they contain genes that code for specific proteins. Ex. Campylobacter Jejuni and VgrG. $\endgroup$ – codax Oct 5 '15 at 23:37
  • $\begingroup$ @codax OK, if you have a list of proteins you want, get their sequences from a related species of bacterium, the closer the better, and use those sequences to query your target genomes. The methods I describe are sensitive enough to deal with small differences in sequence. $\endgroup$ – terdon Oct 6 '15 at 10:37
  • $\begingroup$ I've tried your approach to this problem. The HSP's indicate an overall similarity to all of the sequences entered into the query. From what I understand, there is a near 100% identity ranging from 97-100%. Now I know that there is a high similarity between the list of proteins entered and the species I've entered. Would I have to put one query at a time and blast it against one species to confirm if the sequences code for the individual proteins? $\endgroup$ – codax Oct 16 '15 at 0:44

Do multiple sequence alignment of your protein sequences. This will give you evolutionary relationship between your sequences.

  • $\begingroup$ I think the OP needs to identify genes in the genomic sequence and only has proteins from one species. Multiple alignment won't help. $\endgroup$ – terdon Oct 1 '15 at 12:34
  • 1
    $\begingroup$ I have to identify if these genes exist in a list of species. I have the names of a few proteins. MSA would not apply in this situation. I can retrieve the FASTA containing the entire genomes of these species. Is there a way that I could retrieve the protein sequences and compare them with the nucleotide sequence somehow? $\endgroup$ – codax Oct 4 '15 at 22:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.