Systematic Artifacts in Support Vector Regression-Based Compound Potency Prediction Revealed by Statistical and Activity Landscape Analysis

Support vector machines are a popular machine learning method for many classification tasks in biology and chemistry. In addition, the support vector regression (SVR) variant is widely used for numerical property predictions. In chemoinformatics and pharmaceutical research, SVR has become the probably most popular approach for modeling of non-linear structure-activity relationships (SARs) and predicting compound potency values. Herein, we have systematically generated and analyzed SVR prediction models for a variety of compound data sets with different SAR characteristics. Although these SVR models were accurate on the basis of global prediction statistics and not prone to overfitting, they were found to consistently mispredict highly potent compounds. Hence, in regions of local SAR discontinuity, SVR prediction models displayed clear limitations. Compared to observed activity landscapes of compound data sets, landscapes generated on the basis of SVR potency predictions were partly flattened and activity cliff information was lost. Taken together, these findings have implications for practical SVR applications. In particular, prospective SVR-based potency predictions should be considered with caution because artificially low predictions are very likely for highly potent candidate compounds, the most important prediction targets.     

Modelling interactions of acid–base balance and respiratory status in the toxicity of metal mixtures in the American oyster Crassostrea virginica

Heavy metals, such as copper, zinc and cadmium, represent some of the most common and serious pollutants in coastal estuaries. In the present study, we used a combination of linear and artificial neural network (ANN) modelling to detect and explore interactions among low-dose mixtures of these heavy metals and their impacts on fundamental physiological processes in tissues of the Eastern oyster, Crassostrea virginica. Animals were exposed to Cd (0.001–0.400 μM), Zn (0.001–3.059 μM) or Cu (0.002–0.787 μM), either alone or in combination for 1 to 27 days. We measured indicators of acid–base balance (hemolymph pH and total CO₂), gas exchange (Po₂), immunocompetence (total hemocyte counts, numbers of invasive bacteria), antioxidant status (glutathione, GSH), oxidative damage (lipid peroxidation; LPx), and metal accumulation in the gill and the hepatopancreas. Linear analysis showed that oxidative membrane damage from tissue accumulation of environmental metals was correlated with impaired acid–base balance in oysters. ANN analysis revealed interactions of metals with hemolymph acid–base chemistry in predicting oxidative damage that were not evident from linear analyses. These results highlight the usefulness of machine learning approaches, such as ANNs, for improving our ability to recognize and understand the effects of sub-acute exposure to contaminant mixtures.     

Trypanosoma cruzi: sequence analysis of the variable region of kinetoplast minicircles

The comparisons of 170 sequences of kinetoplast DNA minicircle hypervariable region obtained from 19 stocks of Trypanosoma cruzi and 2 stocks of Trypanosoma cruzi marenkellei showed that only 56% exhibited a significant homology one with other sequences. These sequences could be grouped into homology classes showing no significant sequence similarity with any other homology group. The 44% remaining sequences thus corresponded to unique sequences in our data set. In the DTU I ("Discrete Typing Units") 51% of the sequences were unique. In contrast, in the DTU IId, 87.5% of sequences were distributed into three classes. The results obtained for T. cruzi marinkellei, showed that all sequences were unique, without any similarity between them and T. cruzi sequences. Analysis of palindromes in all sequence sets show high frequency of the EcoRI site. Analysis of repetitive sequences suggested a common ancestral origin of the kDNA. The editing mechanism that occurs in kinetoplastidae is discussed.     

Unlocking the barley genome by chromosomal and comparative genomics

We used a novel approach that incorporated chromosome sorting, next-generation sequencing, array hybridization, and systematic exploitation of conserved synteny with model grasses to assign ~86% of the estimated ~32,000 barley (Hordeum vulgare) genes to individual chromosome arms. Using a series of bioinformatically constructed genome zippers that integrate gene indices of rice (Oryza sativa), sorghum (Sorghum bicolor), and Brachypodium distachyon in a conserved synteny model, we were able to assemble 21,766 barley genes in a putative linear order. We show that the barley (H) genome displays a mosaic of structural similarity to hexaploid bread wheat (Triticum aestivum) A, B, and D subgenomes and that orthologous genes in different grasses exhibit signatures of positive selection in different lineages. We present an ordered, information-rich scaffold of the barley genome that provides a valuable and robust framework for the development of novel strategies in cereal breeding.     

Biochemical and physiological mechanisms related to cold acclimation and enhanced freezing tolerance in poplar plantlets

Temperature is one of the abiotic factors limiting growth and productivity of plants. In the present work, the effect of low non-freezing temperature, as inducer of 'cold acclimation', was studied in poplar. Actively growing plantlets of Populus tremula  ×  Populus tremuloides cv. Muhs 1 were used, and cold treatment consisted in whole plants exposure to 4°C in controlled conditions. Leaves of cold-treated poplars were shown to be acclimated, as an increase of their freezing tolerance was measured using electrolyte leakage. Chlorophyll fluorescence measurements revealed a decrease in photosystem II efficiency while the pigment contents of leaves did not vary. In contrast, after 1 week of cold exposure, an accumulation of pigments was noted in the stems near the apex of the stressed plants as confirmed by chromatographic analyses. Simultaneously, a rapid accumulation of osmoprotectants, i.e. carbohydrates (measured by spectrometry), and of stress indicators (e.g. putrescine) occurred; changes in protein patterns also arose. Indeed, Western blot studies revealed that the expression of three families of stress-related proteins, i.e. dehydrins, stress protein 1 and heat-shock protein 70, was activated or induced by low temperatures. This study complements a previous work on proteomic and individual carbohydrates and provides insight in the ability of poplar plantlets to cold acclimate and to cope with low temperatures by diverse mechanisms (growth cessation, carbohydrate, pigment, polyamine and protein accumulations) related to stress response or involved in acclimation process.     

Synthesis of linear and cyclic phosphopeptides as ligands for the N-terminal SH2-domain of protein tyrosine phosphatase SHP-1.

Linear and cyclic phosphopeptides related to the pY2267 binding site of the epithelial receptor tyrosine kinase Ros have been synthesized as ligands for the amino-terminal SH2 (src homology) domain of protein tyrosine phosphatase SHP-1. The synthesis was accomplished by Fmoc-based solid-phase methodology using side-chain unprotected phosphotyrosine for the linear and mono-benzyl protected phosphotyrosine for the cyclic peptides. According to molecular modelling, the incorporation of a glycine residue between Lys (position pY-1 relative to phosphotyrosine) and Asp or Glu (position pY+2) was recommended for the cyclic candidates. The preparation of these peptides was successfully performed by the incorporation of a Fmoc-Xxx(Gly-OAll)-OH (Xxx = Asp, Glu) dipeptide building block that was prepared in solution prior to SPPS. The cyclization was achieved with PyBOP following Alloc/OAll-deprotection. This study demonstrates the usefulness of allyl-type protecting groups for the generation of side-chain cyclized phosphopeptides. Alloc/OAll-deprotection and cyclization are compatible with phosphorylated tyrosine.     

The membrane form of the DNA repair protein Ku interacts at the cell surface with metalloproteinase 9

The Ku heterodimer (Ku70/Ku80) plays a central role in DNA double-strand breaks repair. Ku is also expressed on the cell surface of different types of cells where its function remains poorly understood. From a yeast two-hybrid screen, we have identified a specific interaction between the core region of Ku80 and the hemopexin domain of metalloproteinase 9 (MMP-9), a key enzyme involved in the degradation of extracellular matrix (ECM) components. Ku associates with MMP-9 on the surface of leukemic cells as demonstrated by co-immunoprecipitation experiments in membrane extracts and double-label immunofluorescence studies. In normal and tumoral migratory cells, Ku80 and MMP-9 colocalize at the periphery of leading edge of cells and cellular invasion of collagen IV matrices was blocked by antibodies directed against Ku70 or Ku80 subunits as well as by Ku80-specific antisense oligonucleotides. Our results indicate that Ku and MMP-9 interact at the cell membrane of highly invasive hematopoietic cells of normal and tumoral origin and document the unexpected importance of the membrane-associated form of Ku in the regulation of ECM remodelling.