Computational Approach For Identifying Gene Families Apart From Similarity Search Based Methods(Reiterative Search Methods)

Question

[chevivien]

Hallo ...i want to identify R gene family in a particular crop..from my literature review some studies has been done on the same using reiterative search process(which involves use of BLAST,....using known genes as query...then find out the conserved domains using commonly know techniques as well as phylogeny analysis.....The number of genes identified are so many and their are chances some results are false positive or some are pseudo genes ...hence i wish to screen the same genes but using different approach altogether....which other approach for gene family finding is more stringent? Also in this particular crop around 500 R (All of them has NBS conserved domain)genes been identified and they are known to confer resistance to various diseases.e.g bacterial diseases,fungal diseases e.t.c ....among the 500 identified genes 20 R genes has been identified to confer Resistance to fungal disease through experimental work....so i wish to screen for more of the R genes conferring resistance to the fungal disease but my problem is how to discriminate it...given that all R genes has conserved domains....what i wish to find is this particular pattern or gene which makes the R genes to be specific to fungal disease but iam not sure how to approach it...any suggestion will be appreciated.

Answer 1

[arnstrm]

For detecting family members, I would recommend generating HMM profiles for each gene families (based on reference genes of the well annotated genomes) and scanning the sequences with those profiles. Once you find the potential members for those profiles, you can reverse blast the sequence to those well annotated genomes to see if they detect the same genes or not. This method in general is more sensitive than blast alone in detecting family members and will work with gene families having high baseline identity OR, If you want to detect more genes that are like 20 R genes, you can make HMM profile with those 20 genes and scan the rest.

[chevivien]

Thanks for your suggestion...helpful indeed...

[Tamador]

could you tell me how I can build HMM profile for detecting family members??

[arnstrm]

Sure, first collect all the sequences for which you want to build the profile, next do an alignment (use MAFFT or ClustalOmega or Muscle or any other aligner of your choice), and then use hmmbuild from the hmmer package to build the profile.

[Tamador]

Thanks , but I need to build the profile without using any software I should write c++ code