Alignment Free Cog Detection Program

Question

[Monzoor]

Is there any composition program available that can classify a given protein sequence to a COG category without the need for alignment?

[Leena]

I too am looking out for such a program. Not for COG finding though. Did someone try to cluster protein sequences using compositional features ?

[Darked89]

re Leena In short yes. See i.e. the paper http://www.pnas.org/content/105/40/15352.short

[Karl]

What's a COG category? Can you elaborate on the problem?

[Torst]

It's unlikely such a program could exist. Something as simple as amino acid composition would not be able to accurately discriminate millions of proteins into a few thousand categories.

[Michael]

It's an old question, so you are possibly not looking anymore, but why would you need an alignment free algorithm?

Answer 1

[Larry_Parnell]

As sequence data are so readily obtainable and available and as alignment methods are so robust, this is indeed the preferred mechanism for COG category assignment. I know of no work to assign a protein to a COG without using an alignment of sequence data.

That said, it is possible to ignore or not use sequence alignments, but it is just not routine.

[Michael]

I have the impression that alignment-free algorithms might be for people who don't believe in evolution lol, so they cannot use it because that would (indirectly) contradict their dogma. For all others, I think they could use the algorithm with best performance.