1. Try using AlienHunter for identifying Islands. This is more easily sold in papers than using a plain GC% cutoff as an identifier.
2. Instead of looking at whole genome alignments as an indicator of evolution, look at each orthologous gene group. For more details on this – look up Martincorena et al – Nature 2012. It involves a series of steps from performing bidirectional BLAST (published by Moreno-Hagelsieb), calculating median pairwise dS / tree dS values, and then removing those families where the ratio is too low.
3. Do that to identify your core genome. I can provide a set of genes that have passed the bidirectional BLAST filter.
4. Do a functional analysis (COG based tests will do) of genes that undergo frequent horizontal replacement.
5. Identify islands that are dynamic. Find islands that are static. What functions do they encode?
6. Four genes for a phylogenetic tree will not suffice. Use the entire Core genome.
7. For gene level phylogeny, use prank (Ari Löytynoja – Science, 2008). Works best for DNA multiple alignment.
The results and its impact doesn’t show up very well. The questions have to be framed more clearly, and the results have to be linked to the questions. The work seems unpublishable because much of what you are saying is not novel. Even then, PLoS One should consider the work. Having said that, there is much that is not known in the system you are studying. Refine your questions in the light of what is already known.
It is often about the depth of inquiry, not its breadth.
A meta-advice: Post it on arXiv (http://arxiv.org) and/or biorXiv (http://biorxiv.org). They are official pre-print servers which are perfect to disseminate such unpublished papers to the community, without them getting lost, and still getting priority.
If funding is a problem you can try to apply for a full submission waiver to a journal. I do research without any funds and it worked beautifully for me in Frontiers.
1. Try using AlienHunter for identifying Islands. This is more easily sold in papers than using a plain GC% cutoff as an identifier.
2. Instead of looking at whole genome alignments as an indicator of evolution, look at each orthologous gene group. For more details on this – look up Martincorena et al – Nature 2012. It involves a series of steps from performing bidirectional BLAST (published by Moreno-Hagelsieb), calculating median pairwise dS / tree dS values, and then removing those families where the ratio is too low.
3. Do that to identify your core genome. I can provide a set of genes that have passed the bidirectional BLAST filter.
4. Do a functional analysis (COG based tests will do) of genes that undergo frequent horizontal replacement.
5. Identify islands that are dynamic. Find islands that are static. What functions do they encode?
6. Four genes for a phylogenetic tree will not suffice. Use the entire Core genome.
7. For gene level phylogeny, use prank (Ari Löytynoja – Science, 2008). Works best for DNA multiple alignment.
The results and its impact doesn’t show up very well. The questions have to be framed more clearly, and the results have to be linked to the questions. The work seems unpublishable because much of what you are saying is not novel. Even then, PLoS One should consider the work. Having said that, there is much that is not known in the system you are studying. Refine your questions in the light of what is already known.
It is often about the depth of inquiry, not its breadth.
I see a lot of you downloaded the paper. It would be very nice of you if you could give me a feedback on the work. That does mean that I am only looking for positive feedback.