The results are a little worse than the two-pass method using sequence-level adaptation, but generally better than the static approach.
This method uses sequence homology between the protein of interest and a protein whose 3D structure is known to predict a three-dimensional model.
The EnzymeDetector [ 6] annotation method uses sequence similarity analysis and a comprehensive enzyme database, BRaunschweig ENzyme DAtabase (BRENDA) [ 7], which is manually extracted from the literature.
For a given drug, our method uses sequence order independent structure alignment, hierarchical clustering and probabilistic sequence similarity to construct a probabilistic pocket ensemble (PPE) that captures promiscuous structural features of different binding sites on known targets.
It is therefore superior to methods using sequence trees, which gave conflicting and debatable results because of tree reconstruction problems [ 1, 13, 116].
In addition to these comparisons with methods using sequence information only, we compared INTREPID to the machine learning algorithms reported by Petrova and Wu (2006) and by Youn et al. (2007) which make use of structural information.
Furthermore, methods using sequence coverage to identify repeats (e.g. Malde et al., 2006; Phillippy et al., 2008) should not be applied to pyrosequencing data without first filtering duplicates.
The PeakPicker ASE method uses sequencing trace data to quantify the relative fluorescent intensity for both alleles at heterozygous SNPs.
Established methods use sequence or structure similarity to infer functions but those types of data do not suffice to determine the biological context in which proteins act.
These methods use sequence templates [ 7] and profiles of the sequences [ 8] as features.
Other methods use sequence quality information derived from the raw trace data.
