Multiple Profile Models Extract Features from Protein Sequence Data and Resolve Functional Diversity of Very Different Protein Families

Abstract

Functional classification of proteins from sequences alone has become a critical bottleneck in understanding the myriad of protein sequences that accumulate in our databases. The great diversity of homologous sequences hides, in many cases, a variety of functional activities that cannot be anticipated. Their identification appears critical for a fundamental understanding of the evolution of living organisms and for biotechnological applications. ProfileView is a sequence-based computational method, designed to functionally classify sets of homologous sequences. It relies on two main ideas: the use of multiple profile models whose construction explores evolutionary information in available databases, and a novel definition of a representation space in which to analyse sequences with multiple profile models combined together. ProfileView classifies protein families by enriching known functional groups with new sequences and discovering new groups and subgroups. We validate ProfileView on seven classes of widespread proteins involved in the interaction with nucleic acids, amino acids and small molecules, and in a large variety of functions and enzymatic reactions. Profile-View agrees with the large set of functional data collected for these proteins from the literature regarding the organisation into functional subgroups and residues that characterise the functions. In addition, ProfileView resolves undefined functional classifications and extracts the molecular determinants underlying protein functional diversity, showing its potential to select sequences towards accurate experimental design and discovery of novel biological functions. On protein families with complex domain architecture, ProfileView functional classification reconciles domain combinations, unlike phylogenetic reconstruction. ProfileView proves to outperform the functional classification approach PANTHER, the two k-mer based methods CUPP and eCAMI and a neural network approach based on Restricted Boltzmann Machines. It overcomes time complexity limitations of the latter.

Unsupervised hierarchical clustering using the learning dynamics of RBMs

“…For instance, in Ref. [35] the authors propose a hierarchical classification method, called ProfileView, that is able to correctly identify 11 of the known subgroups of the CPF protein beyond the capabilities of the phylogenetic tree. It is therefore interesting to compare the tree obtained with our method with the phylogenetic one and the tree obtained with ProfileView, for which we remand to the main article [35].…”

“…[35] the authors propose a hierarchical classification method, called ProfileView, that is able to correctly identify 11 of the known subgroups of the CPF protein beyond the capabilities of the phylogenetic tree. It is therefore interesting to compare the tree obtained with our method with the phylogenetic one and the tree obtained with ProfileView, for which we remand to the main article [35]. As opposed to the phylogenetic tree, our construction is able to readily identify and separate circadian transcriptional regulators from 6-4 photolyases, although it is not able to separate the animal photoreceptors from the 6-4 photolyases as clearly as it happens with ProfileView.…”

“…As opposed to the phylogenetic tree, our construction is able to readily identify and separate circadian transcriptional regulators from 6-4 photolyases, although it is not able to separate the animal photoreceptors from the 6-4 photolyases as clearly as it happens with ProfileView. Also, in [35] the photoreceptors are further separated into the plant and plant-like subgroups. The RBM was able to capture the difference between the two categories, although the plant-like sequences are found in two distinct branches of the tree.…”

Light-driven processes: key players of the functional biodiversity in microalgae

“…Since that discovery, CPF1 with DNA repair activity and diverse light sensing properties have been found in other microalgae, including Chlamydomonas and O. tauri [37], suggesting that a single protein with dual functions might be important in dynamic aquatic environments. Interesting novel CPFs have recently been identified by using a new computational method that extracts functional features from large protein sequence datasets [38].…”

“…Ces protéines présentant une double fonction pourraient donc être particulièrement adaptées aux environnements aquatiques dynamiques. De nouvelles classes de CPF ont récemment été identifiées à l'aide d'une nouvelle méthode de calcul qui extrait les caractéristiques fonctionnelles de grands ensembles de données de séquence protéique [38]. La fonction exacte de ces nouveaux CPF pourra être testée expérimentalement chez des espèces modèles de microalgues pour confirmer qu'il s'agit bien de protéines photosensibles présentant de nouvelles caractéristiques.…”