Adding Stochastic Negative Examples into Machine Learning Improves Molecular Bioactivity Prediction.

Multitask deep neural networks learn to predict ligand-target binding by example, yet public pharmacological data sets are sparse, imbalanced, and approximate. We constructed two hold-out benchmarks to approximate temporal and drug-screening test scenarios, whose characteristics differ from a random split of conventional training data sets. We developed a pharmacological data set augmentation procedure, Stochastic Negative Addition (SNA), which randomly assigns untested molecule-target pairs as transient negative examples during training. Under the SNA procedure, drug-screening benchmark performance increases from R2 = 0.1926 ± 0.0186 to 0.4269 ± 0.0272 (122%). This gain was accompanied by a modest decrease in the temporal benchmark (13%). SNA increases in drug-screening performance were consistent for classification and regression tasks and outperformed y-randomized controls. Our results highlight where data and feature uncertainty may be problematic and how leveraging uncertainty into training improves predictions of drug-target relationships.

Patients with Parkinson's Disease Show Alteration in their Visuospatial Abilities and in their Egocentric and Allocentric Spatial Orientation Measured by Card Placing Tests.

BACKGROUND: Visuospatial skills are impaired in Parkinson's disease (PD). Other related skills exist, such as spatial orientation have been poorly studied. The egocentric (based on internal cues) and allocentric frameworks (based on external cues) are used in daily spatial orientation. Depending on PD onset, the allocentric framework may have a higher level of impairment in tremor-dominant and the egocentric one in akinetic-rigid. OBJECTIVE: To evaluate spatial orientation and visuospatial functions in PD patients and controls, and to assess whether their performance is related to disease duration and the PD subtype (tremor-dominant and akinetic-rigid). METHODS: We evaluated egocentric and allocentric spatial orientation (Egocentric and Allocentric Spatial Memory Tasks) and visuospatial abilities, span and working memory in 59 PD patients and 51 healthy controls. RESULTS: Visuospatial skills, visuospatial span, and egocentric and allocentric orientation are affected in PD. Visuospatial skills and allocentric orientation undergo deterioration during the first 5 years of the disease progression, while egocentric orientation and visuospatial span do so at later stages (9-11 years). The akinetic-rigid subtype presents worse results in all the spatial abilities that were measured when compared to controls, and worse scores in visuospatial working memory, visuospatial abilities and allocentric orientation when compared to the tremor-dominant group. The tremor-dominant group performed worse than controls in egocentric and allocentric orientation. CONCLUSION: PD patients show deficits in their visuospatial abilities and in their egocentric and allocentric spatial orientation compared to controls, specifically in akinetic-rigid PD. Only spatial orientation are affected in tremor-dominant PD patients. Allocentric orientation is affected earlier in the progression of the disease.

Aerobic interval exercise improves renal functionality and affects mineral metabolism in obese Zucker rats.

Obesity, metabolic syndrome, and renal injury are considered risk factors for type 2 diabetes, as well as kidney disease. Functional and structural changes in the kidney as consequence of obesity and metabolic syndrome may lead to impaired mineral metabolism in what is known as chronic kidney disease-mineral and bone disorder. Lifestyle interventions such as physical activity are good strategies to manage these pathologies and therefore, prevent the loss of kidney functionality and related complications in mineral metabolism. In this study, we have used 40 male Zucker rats that were randomly allocated into four different experimental groups, two of them (an obese and a lean one) performed an aerobic interval training protocol, and the other two groups were sedentary. At the end of the experimental period (8 wk), urine, plasma, and femur were collected for biochemical and mineral composition analysis, whereas the kidney was processed for histological studies. The obese rats exhibited albuminuria, glomerulosclerosis, and hypertrophy in glomeruli and renal tubule in some areas, together with alterations in mineral content of plasma but not of femur. The training protocol prevented the generation of albuminuria and glomerulosclerosis, showing a significant action on plasma and bone mineral levels. Therefore, the specific training protocol used in this study was able to prevent the development of diabetic nephropathy and affected the metabolism of certain minerals.

A Simple Representation of Three-Dimensional Molecular Structure.

Statistical and machine learning approaches predict drug-to-target relationships from 2D small-molecule topology patterns. One might expect 3D information to improve these calculations. Here we apply the logic of the extended connectivity fingerprint (ECFP) to develop a rapid, alignment-invariant 3D representation of molecular conformers, the extended three-dimensional fingerprint (E3FP). By integrating E3FP with the similarity ensemble approach (SEA), we achieve higher precision-recall performance relative to SEA with ECFP on ChEMBL20 and equivalent receiver operating characteristic performance. We identify classes of molecules for which E3FP is a better predictor of similarity in bioactivity than is ECFP. Finally, we report novel drug-to-target binding predictions inaccessible by 2D fingerprints and confirm three of them experimentally with ligand efficiencies from 0.442-0.637 kcal/mol/heavy atom.

Virgin Beta Cells Persist throughout Life at a Neogenic Niche within Pancreatic Islets.

Postnatal maintenance or regeneration of pancreatic beta cells is considered to occur exclusively via the replication of existing beta cells, but clinically meaningful restoration of human beta cell mass by proliferation has never been achieved. We discovered a population of immature beta cells that is present throughout life and forms from non-beta precursors at a specialized micro-environment or "neogenic niche" at the islet periphery. These cells express insulin, but lack other key beta cell markers, and are transcriptionally immature, incapable of sensing glucose, and unable to support calcium influx. They constitute an intermediate stage in the transdifferentiation of alpha cells to cells that are functionally indistinguishable from conventional beta cells. We thus identified a lifelong source of new beta cells at a specialized site within healthy islets. By comparing co-existing immature and mature beta cells within healthy islets, we stand to learn how to mature insulin-expressing cells into functional beta cells.

Determination of ubiquitin fitness landscapes under different chemical stresses in a classroom setting.

Ubiquitin is essential for eukaryotic life and varies in only 3 amino acid positions between yeast and humans. However, recent deep sequencing studies indicate that ubiquitin is highly tolerant to single mutations. We hypothesized that this tolerance would be reduced by chemically induced physiologic perturbations. To test this hypothesis, a class of first year UCSF graduate students employed deep mutational scanning to determine the fitness landscape of all possible single residue mutations in the presence of five different small molecule perturbations. These perturbations uncover 'shared sensitized positions' localized to areas around the hydrophobic patch and the C-terminus. In addition, we identified perturbation specific effects such as a sensitization of His68 in HU and a tolerance to mutation at Lys63 in DTT. Our data show how chemical stresses can reduce buffering effects in the ubiquitin proteasome system. Finally, this study demonstrates the potential of lab-based interdisciplinary graduate curriculum.

Urocortin3 mediates somatostatin-dependent negative feedback control of insulin secretion.

The peptide hormone urocortin3 (Ucn3) is abundantly expressed by mature beta cells, yet its physiological role is unknown. Here we demonstrate that Ucn3 is stored and co-released with insulin and potentiates glucose-stimulated somatostatin secretion via cognate receptors on delta cells. Further, we found that islets lacking endogenous Ucn3 have fewer delta cells, reduced somatostatin content, impaired somatostatin secretion, and exaggerated insulin release, and that these defects are rectified by treatment with synthetic Ucn3 in vitro. Our observations indicate that the paracrine actions of Ucn3 activate a negative feedback loop that promotes somatostatin release to ensure the timely reduction of insulin secretion upon normalization of plasma glucose. Moreover, Ucn3 is markedly depleted from beta cells in mouse and macaque models of diabetes and in human diabetic islets. This suggests that Ucn3 is a key contributor to stable glycemic control, whose reduction during diabetes aggravates glycemic volatility and contributes to the pathophysiology of this disease.

The transcriptional landscape of mouse beta cells compared to human beta cells reveals notable species differences in long non-coding RNA and protein-coding gene expression.

BACKGROUND: Insulin producing beta cell and glucagon producing alpha cells are colocalized in pancreatic islets in an arrangement that facilitates the coordinated release of the two principal hormones that regulate glucose homeostasis and prevent both hypoglycemia and diabetes. However, this intricate organization has also complicated the determination of the cellular source(s) of the expression of genes that are detected in the islet. This reflects a significant gap in our understanding of mouse islet physiology, which reduces the effectiveness by which mice model human islet disease. RESULTS: To overcome this challenge, we generated a bitransgenic reporter mouse that faithfully labels all beta and alpha cells in mouse islets to enable FACS-based purification and the generation of comprehensive transcriptomes of both populations. This facilitates systematic comparison across thousands of genes between the two major endocrine cell types of the islets of Langerhans whose principal hormones are of cardinal importance for glucose homeostasis. Our data leveraged against similar data for human beta cells reveal a core common beta cell transcriptome of 9900+ genes. Against the backdrop of overall similar beta cell transcriptomes, we describe marked differences in the repertoire of receptors and long non-coding RNAs between mouse and human beta cells. CONCLUSIONS: The comprehensive mouse alpha and beta cell transcriptomes complemented by the comparison of the global (dis)similarities between mouse and human beta cells represent invaluable resources to boost the accuracy by which rodent models offer guidance in finding cures for human diabetes.