Structure Prediction of Partial-Length Protein Sequences

Abstract

Protein structure information is essential to understand protein function. Computational methods to accurately predict protein structure from the sequence have primarily been evaluated on protein sequences representing full-length native proteins. Here, we demonstrate that top-performing structure prediction methods can accurately predict the partial structures of proteins encoded by sequences that contain approximately 50% or more of the full-length protein sequence. We hypothesize that structure prediction may be useful for predicting functions of proteins whose corresponding genes are mapped expressed sequence tags (ESTs) that encode partial-length amino acid sequences. Additionally, we identify a confidence score representing the quality of a predicted structure as a useful means of predicting the likelihood that an arbitrary polypeptide sequence represents a portion of a foldable protein sequence (“foldability”). This work has ramifications for the prediction of protein structure with limited or noisy sequence information, as well as genome annotation.

Characterization and Structure Prediction of Partial Length Protein Sequences of pcoA, pcoR and chrB Genes from Heavy Metal Resistant Bacteria from the Klip River, South Africa

“…The tool that was used in this study to predict the protein structures of the partial proteins was I-TASSER, since it was reported to reliably predict partial structures of a source protein [ 23 ]. I-TASSER initiates the structure prediction process by performing threading to identify template Protein Database Structures that are similar to the query sequence.…”

“…I-TASSER initiates the structure prediction process by performing threading to identify template Protein Database Structures that are similar to the query sequence. Structural fragments that are cut from the chosen templates are then used to construct the protein models [ 23 ]. The C-score output of I-TASSER was used to distinguish the foldable and non-foldable sequences because it is founded on the probability that the sequence is homologous to known structures at the subsequence level rather than globally.…”

“…The C-score output of I-TASSER was used to distinguish the foldable and non-foldable sequences because it is founded on the probability that the sequence is homologous to known structures at the subsequence level rather than globally. Structure predictions with a TM score greater than 0.6 was defined as “good”, “decent” (TM-score between 0.3 and 0.6) or “bad” (TM-scores less than 0.3) [ 23 ].…”

“…Given an arbitrary protein sequence, it would be informative to accurately predict the probability that the partial protein sequence represents a foldable protein sequence. This information would be valuable in genome annotation by supporting or rejecting hypothetical proteins stemming from unfamiliar regions of DNA [ 23 ]. Having an accurately predicted structure of an uncharacterized partial protein sequence is likely to provide information about its function that a sequence alone cannot provide [ 24 , 25 ].…”

A protein sequence fitness function for identifying natural and nonnatural proteins

“…A primary and a very crucial step of novel protein design involves the computational identification or generation of potential protein sequences having a considerably high probability of mimicking the naturally occurring proteins and eventually folding into a compact structure 18,19 . Some of the earlier studies measured the degree of randomness in the known protein sequences to explore the logical explanations with conflicting inferences for constrained available sequence space 5,7,8,19‐22 . Some of the previous computational studies of random sequence proteins argued over the extent of variability among natural protein sequences and random protein sequences 23‐25 .…”