I want a consensus annotation for group of orthologous protein. I know Interpro, KOG/COG, Blast2Go are the nice option to do functional annotation but here I do not always have the similar annotation of a cluster. For, example I have have clusters generated from multiparanoid, I extracted the sequences belong to cluster1 and then did the annotation by these programs but in some cases the annotation is different. So, I want a consensus annotation for each cluster like cluster1 is laminin protein, cluster2 is toll and so on.
Dear Ashmit,
You may take a look at "Gene Family" function implemented on our LegumeIP* - An Integrative Platform to Study Gene Function and Genome Evolution in Legumes (http://plantgrn.noble.org/LegumeIP/). Hope it gives you some ideas with regard to annotating your orthologous clusters.
* Jun Li,Xinbin Dai,Tingsong Liu and Patrick Xuechun Zhao. LegumeIP: an integrative database for comparative genomics and transcriptomics of model legumes. Nucleic Acids Research 2012 Jan ;40(Database issue):D1221-D1229.
All the best,
Patrick
Dear Ashmit
Reaching consensus is certainly a challenging problem. Instead, I suggest you apply for classification features across the tools that you aptly describe and use. The paranoid gene families that you obtain from clusters of orthologous groups can be transferred to batch entrez in GenBank and genecards.org. I would still assume your data to be comparatively smaller.
Now comes the two important questions that you may ask:
1. How similar or different are orthologous clusters?
2. Can we at some point of time find a strategy to predict the outcome of those similar or dissimilar clusters
(a) using extended functional annotation based on motifs, transcribed genes, expression profiles or subcellular location to an extent.
(b) Train a network of genes using machine learning based predictions so as to reach consensus.
Most of the time point 2(a) may not be feasible and you may have to go for the other option.
I hope this helps
Prash
Ashmit,
While I understand your desire to have a simple annotation of gene clusters, protein biology is not that simple and the methods of functional annotation are -I would say- of medium level reliability. As such I would recommend that you lower your expectations of how 'clean' your functional annotation can be, and consider running a functional annotation of all sequences, then analysing across orthologous clusters, shared domains (which are often the functional units anyway), and grouped by function itself.
I recommend dcGO:
http://supfam.org/SUPERFAMILY/dcGO
But don't take my word for it, the leading article on the state of the art is from the CAFA competition (Critical Assessment of Functional Annotation):
http://www.nature.com/nmeth/journal/v10/n3/full/nmeth.2340.html
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools