FROST: A filter‐based fold recognition method

Abstract

To assess the reliability of fold assignments to protein sequences, we developed a fold recognition method called FROST (Fold Recognition-Oriented Search Tool) based on a series of filters and a database specifically designed as a benchmark for this new method under realistic conditions. This benchmark database consists of proteins for which there exists, at least, another protein with an extensively similar 3D structure in a database of representative 3D structures (i.e., more than 65% of the residues in both proteins can be structurally aligned). Because the testing of our method must be carried out under conditions similar to those of real fold recognition experiments, no protein pair with sequence similarity detectable using standard sequence comparison methods such as FASTA is included in the benchmark database. While using FROST, we achieved a coverage of 60% for a rate of error of 1%. To obtain a baseline for our method, we used PSI-BLAST and 3D-PSSM. Under the same conditions, for a 1% error rate, coverages for PSI-BLAST and 3D-PSSM were 33 and 56%, respectively.

Recent Advances in Solving the Protein Threading Problem

“…FROST is intended to assess the reliability of fold assignments to a given protein sequence (hereafter called a query sequence or query for short) [33,7]. To perform this task, FROST used a series of filters, each one possessing a specific scoring function that measures the fitness of the query sequence for template cores.…”

“…A description of the implementation of these different components in the FROST (Fold Recognition Oriented Search Tool) method [7] is detailed in the next section. Let us note that, from a computer scientist's viewpoint, the third component above is the most challenging part of the treading method development.…”

“…Two protein threading packages are currently available that implement exact methods based on the latter approach: RAPTOR 5 [12] and FROST 6 [7]. In section 3 we will describe in more details the FROST implementation of the MIP models.…”

“…The assumption behind this procedure is that when a sequence is compatible with a core, its length must be similar to the core length (± 30%). 7 We align the sequences of each set with the core. This provides empirical distributions of scores for aligning sequences with different lengths onto the core.…”

“…Unfortunately, since we do not know the shape nor the parameters of the distributions, we cannot compute a precise probability for the sequence to belong to this population of unrelated sequences. We use empirical results obtained on a test database to estimate when a normalized score is significant at a 99% level of confidence [33,7] (see next section).…”