gene homology evidence

Question

[mina]

Hello everyone

I have tried to find homologous of particular genes of Arabidopsis in Transcriptome data (another plant). I did tblastn specific gene of Arabidopsis against this data. The result was E-vaue :6e-102 with 88% Identity. So this 2 genes may be homologs. I did CDC(conserve domain), MSA (multiple seq alignment) but my supervisor told me that you should bring more evidence to prove that this 2 genes are homologs. what can I do more to show they are homologs? I only have fasta format of transcriptome and Protein seq.

thank you in forward

regards

Answer 1

[Siva]

Sequence similarity between two sequences alone does not imply homology. There could be other reasons for sequence similarity between two sequences (convergent evolution, horizontal gene transfer etc.,) though this is rare in eukaryotes. If you want to use sequence similarity, see if these two sequences are bi-directional best hits (BBH). You need to do phylogenetic analyses to be more confident about homology.

[mina]

Thank you so much Siva. This helps me lot.

So, if top reciprocal BLAST hit is my gene of interest, these 2 genes (between particular plant and Arabidopsis) are homologous. Is that right?

In addition , in the case of phylogenetic analysis, you mean phylogenetic tree? Can I use the BLASTX top 100 hits to draw the phylogenetic tree?

And in phylogenetic tree, how far these two genes could be (from each other), to call them homologous? Or they just have a last comment ancestor, is enough?

Which software would you recommend for phylogenetic tree? By neighbor joining method or not?

English is not my first language, so please excuse any mistakes.

Thanks in forward.

Regards

[Siva]

So, if top reciprocal BLAST hit is my gene of interest , these 2 genes (between particular plant and Arabidopsis) are homologous. Is that right?

Sort of. For most practical purposes, two sequences with each being their BBHs with an alignment length of >80% should be enough to call them putative orthologs. Since you already did the multiple alignment with similar proteins, check for domain organization, conserved residues etc.,. to strengthen your hypothesis that these two sequences are putative orthologs.

In addition, in the case of phylogenetic analysis, you mean phylogenetic tree? Can I use the BLASTX top 100 hits to draw the phylogenetic tree?

Yes, building phylogenetic trees. It depends on what you have in the top 100 hits. You want to have sequences from diverse lineages to build tree. I would take as many as few hundred hits and cluster them to remove highly redundant sequences before building the tree.

And in phylogenetic tree, how far these two genes could be (from each other), to call them homologous? Or they just have a last comment ancestor, is enough?

Please check this excellent review/tutorial to understand building and interpreting trees.

Which software would you recommend for phylogenetic tree? By neighbor joining method or not?

Use maximum likelihood methods implemented in PhyML or RAxML. If you don't want to install these software, you can use the Phylogeny.fr website to run them online.