Faster homology detection using sequence profiles

In an attempt to make homology detection faster, scalable and sensitive, both the natural and designed sequences were represented as hidden Markov models. These profiles capture the evolutionary information of protein sequences and result in accurate homology detection.

There was an appreciable improvement in time as compared to our previous work. It is recommended to use multiple methods in concert for homology driven fold assignments.

The assessment on fold associated protein families showed promise in applying the method on structure unknown protein families.

The caveat that remained was the limitation of providing fold associations for multi-domain protein families.

Using these sequence profiles, fold recognition for 1314 protein families were provided. For 20 protein families, we associated the correct fold for 19.