Performance of secondary structure prediction methods on proteins containing structurally ambivalent sequence fragments

Abstract

Several approaches for predicting secondary structures from sequences have been developed and reached a fair accuracy. One of the most rigorous tests for these prediction methods is their ability to correctly predict identical fragments of protein sequences adopting different secondary structures in unrelated proteins. In our previous work, we obtained 30 identical octapeptide sequence fragments adopting different backbone conformations. It is of interest to find whether the presence of structurally ambivalent fragments in proteins will affect the accuracy of secondary structure prediction methods or not. Hence, in this work, we have made a systematic comparative analysis on secondary structure prediction results of 30 identical octapeptide pairs and 52 identical heptapeptide pairs adopting different conformations with the aid of segment overlap measure. The results reveal the better performance of profile-based methods such as PSIpred and JPred and misprediction by classical rule-based methods such as Garnier Osguthorpe Robson Method and Double Prediction Method. The results discussed here insist that modern secondary structure prediction methods are able to better discriminate conformationally ambivalent peptide fragments.

Comprehensive analysis of phenotypes and genetics in 21 Chinese families with haemophilia B: characterization of five novel mutations

“…One patient had a novel in‐frame duplication of 12 nucleotides (TATATACCAAGG) at codon 448_451, which encodes four additional amino acids (Tyr, Thr, Lys, and Val) in subdomain‐2 (residues 334–461) of the catalytic domain. The secondary structure of the in‐frame duplication of p.448_451DupYTKV analysed using the method of PSIPRED was changed to a coil from the beta‐strand in wild‐type FIX (Fig. ).…”

Turncoat Polypeptides: We Adapt to Our Environment

“…Finally, third‐generation, advanced algorithms benefit from the increase in elucidated three‐dimensional structures, make use of evolutionary information through conservation analyses of multiple sequence alignments, and incorporate long‐range interactions in their calculations. Therefore, it is hardly surprising that the structure of chameleon sequences in their native proteins can be mostly accurately predicted by using these profile‐based methods, even with similar accuracy to the rest of the sequences . However, for the same reason, they are of limited use to discover fold‐switching regions.…”

“…Therefore, it is hardly surprising that the structure of chameleon sequences in their native proteinsc an be mostly accurately predicted by using these profile-based methods, even with similar accuracyt ot he rest of the sequences. [10,33,[43][44][45] However, for the same reason, they are of limited use to discover fold-switching regions. On this basis, Porter and Loogerd eveloped as earching procedure that employed "incorrect" secondary structure predictions( carriedo ut with SPIDER 2; [46] this method does not employ long-range sequencei nformation), together with the estimation of independent folding cooperativity.…”

“…Regarding amino acid composition, hydrophobic residues, such as leucine, isoleucine, valine, and alanine, are found with high frequency in chameleon sequences. [35,45] Moreover,t hese structures contain am ix of amino acids that strongly favor both a-helix and b-strand conformations, [33,45,47] whereas secondary-structure breakers, such as proline, are, in turn, underrepresented. [45]…”

Predicting Switch-Like Behavior in Proteins Using Logistic Regression on Sequence-Based Descriptors

“…The variable 1 is the number of times that the most common type of secondary structure occurs in all of the predictions made. For our purposes, = 15, since we use the 15 algorithms for calculating the prediction of secondary structure originally applied by Saravanan and Selvaraj [16]. The minimum and maximum values for…”