How to get best predictions for road monitoring using machine learning techniques.

Road condition monitoring is essential for improving traffic safety and reducing accidents. Machine learning methods have recently gained prominence in the practically important task of controlling road surface quality. Several systems have been proposed using sensors, especially accelerometers present in smartphones due to their availability and low cost. However, these methods require practitioners to specify an exact set of features from all the sensors to provide more accurate results, including the time, frequency, and wavelet-domain signal features. It is important to know the effect of these features change on machine learning model performance in handling road anomalies classification tasks. Thus, we address such a problem by conducting a sensitivity analysis of three machine learning models which are Support Vector Machine, Decision Tree, and Multi-Layer Perceptron to test the effectiveness of the model by selecting features. We built a feature vector from all three axes of the sensors that boosts classification performance. Our proposed approach achieved an overall accuracy of 94% on four types of road anomalies. To allow an objective analysis of different features, we used available accelerometer datasets. Our objective is to achieve a good classification performance of road anomalies by distinguishing between significant and relatively insignificant features. Our chosen baseline machine learning models are based on their comparative simplicity and powerful empirical performance. The extensive analysis results of our study provide practical advice for practitioners wishing to select features effectively in real-world settings for road anomalies detection.

Two distinct regions of calponin share common binding sites on actin resulting in different modes of calponin-actin interaction.

Calponins are a small family of proteins that alter the interaction between actin and myosin II and mediate signal transduction. These proteins bind F-actin in a complex manner that depends on a variety of parameters such as stoichiometry and ionic strength. Calponin binds G-actin and F-actin, bundling the latter primarily through two distinct and adjacent binding sites (ABS1 and ABS2). Calponin binds other proteins that bind F-actin and considerable disagreements exist as to how calponin is located on the filament, especially in the presence of other proteins. A study (Galkin, V.E., Orlova, A., Fattoum, A., Walsh, M.P. and Egelman, E.H. (2006) J. Mol. Biol. 359, 478-485.), using EM single-particle reconstruction has shown that there may be four modes of interaction, but how these occur is not yet known. We report that two distinct regions of calponin are capable of binding some of the same sites on actin (such as 18-28 and 360-372 in subdomain 1). This accounts for the finding that calponin binds the filament with different apparent geometries. We suggest that the four modes of filament binding account for differences in stoichiometry and that these, in turn, arise from differential binding of the two calponin regions to actin. It is likely that the modes of binding are reciprocally influenced by other actin-binding proteins since members of the alpha-actinin group also adopt different actin-binding positions and bind actin principally through a domain that is similar to calponin's ABS1.

The gelsolin:calponin complex nucleates actin filaments with distinct morphologies.

Gelsolin and calponin are cytoskeletal and signalling proteins that form a tight 1:1 complex (GCC). We show that calponin within the GCC inhibits the rate of gelsolin mediated nucleation of actin polymerization. The actin-binding function of calponin is ablated within the GCC as the actin-binding site overlaps with one of the gelsolin binding sites. The structure of filaments that result from nucleation by GCC are different to those nucleated by gelsolin alone in that they are longer, loosely bundled and stain heterogeneously with phalloidin. GCC nucleated filaments appear contorted and wrap around each to form the loose bundles.

Two distinct sites of interaction form the calponin: gelsolin complex and two calcium switches control its activity.

Gelsolin and calponin are well characterized actin-binding proteins that form a tight gelsolin:calponin complex (GCC). We show here that the GCC is formed through two distinct interfaces. One of these is formed between 144-182 of calponin and 25-150 of gelsolin (G1). The second is a calcium-sensitive site centred on calponin's CH domain, and the C-terminal half of gelsolin (G4-6). The behaviour of this second interface is dependent on the presence of calcium and so it is possible that potential GCC-binding partners may be selected by calcium availability. Actin is one such GCC-binding partner and we show that a larger complex is formed with monomeric actin in calcium. The stoichiometry of this complex is determined to be 1 gelsolin/1 calponin/2 G-actins (GCA(2)). Both actin monomers bind the GCC through gelsolin. Both calponin and gelsolin are reported to play signaling roles in addition to their better-characterized actin-binding properties and it is possible that the GCC regulates both of these functions.

Calcium-induced conformational changes in the amino-terminal half of gelsolin.

Gelsolin is an actin-binding protein that is regulated by the occupancy of multiple calcium-binding sites. We have studied calcium induced conformational changes in the G1-2 and G1-3 sub-domains, and report the binding affinities for the three type II sites. A new probe for G3 has been produced and a K(d) of 5 microM has been measured for calcium in the context of G1-3. The two halves of gelsolin, G1-3 and G4-6 bind weakly with or without calcium, suggesting that once separated by apoptotic proteolysis, G1-3 and G4-6 remain apart allowing G1-3 to sever actin in a calcium free manner.

Calponin binds G-actin and F-actin with similar affinity.

Calponins are actin-binding proteins that are implicated in the regulation of actomyosin. Calponin binds filamentous actin (F-actin) through two distinct sites ABS1 and ABS2, with an affinity in the low micromolar range. We report that smooth muscle calponin binds monomeric actin with a similar affinity (K(d) of 0.15 microM). We show that the arrangement of binding is similar to that of F-actin by a number of criteria, most notably that the distance between Cys273 on calponin and Cys374 of actin is 29A when measured by fluorescent resonance energy transfer, the same distance as previously reported for F-actin.

A direct interaction with calponin inhibits the actin-nucleating activity of gelsolin.

Gelsolin and calponin are well-characterized cytoskeletal proteins that are abundant and widely expressed in vertebrate tissues. It is also becoming apparent, however, that they are involved in cell signalling. In the present study, we show that gelsolin and calponin interact directly to form a high-affinity (K(d)=16 nM) 1:1 complex, by the use of fluorescent probes attached to both proteins, by affinity chromatography and by immunoprecipitation. These methods show that gelsolin can form high-affinity complexes with two calponin isoforms (basic h1 and acidic h3). They also show that gelsolin binds calponin through regions that have been identified previously as being calponin's actin-binding sites. Moreover, gelsolin does not interact with calponin while calponin is bound to F-actin. Reciprocal experiments to find calponin-binding sites on gelsolin show that these are in both the N- and C-terminal halves of gelsolin. Calponin has minimal effects on actin severing by gelsolin. In contrast, calponin markedly affects the nucleation activity of gelsolin. The maximum inhibition of nucleation by gelsolin was 50%, which was achieved with a ratio of two calponins for every gelsolin. Thus the interaction of calponin with gelsolin may play a regulatory role in the formation of actin filaments through modulation of gelsolin's actin-binding function and through the prevention of calponin's actin-binding activities.