Computationally Efficient Demographic History Inference from Allele Frequencies with Supervised Machine Learning.

Inferring past demographic history of natural populations from genomic data is of central concern in many studies across research fields. Previously, our group had developed dadi, a widely used demographic history inference method based on the allele frequency spectrum (AFS) and maximum composite-likelihood optimization. However, dadi's optimization procedure can be computationally expensive. Here, we present donni (demography optimization via neural network inference), a new inference method based on dadi that is more efficient while maintaining comparable inference accuracy. For each dadi-supported demographic model, donni simulates the expected AFS for a range of model parameters then trains a set of Mean Variance Estimation neural networks using the simulated AFS. Trained networks can then be used to instantaneously infer the model parameters from future genomic data summarized by an AFS. We demonstrate that for many demographic models, donni can infer some parameters, such as population size changes, very well and other parameters, such as migration rates and times of demographic events, fairly well. Importantly, donni provides both parameter and confidence interval estimates from input AFS with accuracy comparable to parameters inferred by dadi's likelihood optimization while bypassing its long and computationally intensive evaluation process. donni's performance demonstrates that supervised machine learning algorithms may be a promising avenue for developing more sustainable and computationally efficient demographic history inference methods.

Computationally efficient demographic history inference from allele frequencies with supervised machine learning.

Inferring past demographic history of natural populations from genomic data is of central concern in many studies across research fields. Previously, our group had developed dadi, a widely used demographic history inference method based on the allele frequency spectrum (AFS) and maximum composite likelihood optimization. However, dadi's optimization procedure can be computationally expensive. Here, we developed donni (demography optimization via neural network inference), a new inference method based on dadi that is more efficient while maintaining comparable inference accuracy. For each dadi-supported demographic model, donni simulates the expected AFS for a range of model parameters then trains a set of Mean Variance Estimation neural networks using the simulated AFS. Trained networks can then be used to instantaneously infer the model parameters from future input data AFS. We demonstrated that for many demographic models, donni can infer some parameters, such as population size changes, very well and other parameters, such as migration rates and times of demographic events, fairly well. Importantly, donni provides both parameter and confidence interval estimates from input AFS with accuracy comparable to parameters inferred by dadi's likelihood optimization while bypassing its long and computationally intensive evaluation process. donni's performance demonstrates that supervised machine learning algorithms may be a promising avenue for developing more sustainable and computationally efficient demographic history inference methods.

A new dexpanthenol-containing liquid cleanser for atopic-prone skin: Results from two prospective clinical studies evaluating cutaneous tolerability, moisturization potential, and effects on barrier function.

BACKGROUND: Gentle cleansing of the skin without further compromising its barrier function and moisture content and being simultaneously devoid of adverse effects on the skin microbiome are important features of body cleansers for atopic-prone skin sufferers. For this population, a new dexpanthenol-containing liquid cleanser (DCLC) was developed. METHODS: Two prospective 4-week studies have been conducted. Study 1 investigated the effect of once-daily DCLC on stratum corneum (SC) hydration, transepidermal water loss (TEWL), skin pH, and skin microbiome (all on the volar forearm) in adult subjects with dry skin (N = 44). Study 2 explored the cutaneous tolerability of DCLC and its effect on the microbiome biodiversity of the volar forearm skin in infants/children with atopic-prone skin (N = 33, aged 6 months to 6 years). In the latter study, DCLC was applied 2-3 days/week in combination with an emollient applied at least twice daily. RESULTS: In Study 1, on Day 29, the mean change in skin surface capacitance from baseline was significantly greater in the forearm test area treated with DCLC than in the contralateral test area (control) treated with water only (5.16 vs. 3.65 a.u.; p = 0.011), suggesting long-term SC hydration. DCLC use was not associated with changes in TEWL, skin pH, or microbiome biodiversity if compared to control. In Study 2, the 4-week use of DCLC in combination with an emollient was well tolerated according to pediatrician skin assessment, and no flare-ups were recorded. The microbiome biodiversity did not shift during the study. CONCLUSION: These findings support the use of DCLC in subjects with atopic-prone skin.

PND-1186 FAK inhibitor selectively promotes tumor cell apoptosis in three-dimensional environments.

Tumor cells can grow in an anchorage-independent manner. This is mediated in part through survival signals that bypass normal growth restraints controlled by integrin cell surface receptors. Focal adhesion kinase (FAK) is a cytoplasmic protein-tyrosine kinase that associates with integrins and modulates various cellular processes including growth, survival, and migration. As increased FAK expression and tyrosine phosphorylation are associated with tumor progression, inhibitors of FAK are being tested for anti-tumor effects. Here, we analyze PND-1186, a substituted pyridine reversible inhibitor of FAK activity with a 50% inhibitory concentration (IC50) of 1.5 nM in vitro. PND-1186 has an IC50 of ~100 nM in breast carcinoma cells as determined by anti-phospho-specific immunoblotting to FAK Tyr-397. PND-1186 did not alter c­Src or p130Cas tyrosine phosphorylation in adherent cells, yet functioned to restrain cell movement. Notably, 1.0 µM PND-1186 (>5-fold above IC50) had limited effects on cell proliferation. However, under non-adherent conditions as spheroids and as colonies in soft agar, 0.1 µM PND-1186 blocked FAK and p130Cas tyrosine phosphorylation, promoted caspase-3 activation, and triggered cell apoptosis. PND-1186 inhibited 4T1 breast carcinoma subcutaneous tumor growth correlated with elevated tumor cell apoptosis and caspase 3 activation. Addition of PND-1186 to the drinking water of mice was well tolerated and inhibited ascites- and peritoneal membrane-associated ovarian carcinoma tumor growth associated with the inhibition of FAK Tyr-397 phosphorylation. Our results with low-level PND-1186 treatment support the conclusion that FAK activity selectively promotes tumor cell survival in three-dimensional environments.

Discovery of 5-substituted 2-amino-4-chloro-8-((4-methoxy-3,5-dimethylpyridin-2-yl)methyl)-7,8-dihydropteridin-6(5H)-ones as potent and selective Hsp90 inhibitors.

A new series of compounds, 5-substituted 2-amino-4-chloro-8-((4-methoxy-3,5-dimethylpyridin-2-yl)methyl)-7,8-dihydropteridin-6(5H)-ones, have been designed and identified as potent and selective inhibitors of Hsp90. These compounds demonstrated nanomolar potency toward both Hsp90-regulated Her2 degradation and the growth of a panel of human tumor cell lines in cell-based assays. High selectivity of these compounds toward Hsp90 was evident given that they did not inhibit a panel of 34 kinases at 10microM. The structure-activity relationship (SAR) of this series is reported here.