Predicting expected progeny difference for marbling score in Angus cattle using artificial neural networks and Bayesian regression models

Background

Artificial neural networks (ANN) mimic the function of the human brain and are capable of performing massively parallel computations for data processing and knowledge representation. ANN can capture nonlinear relationships between predictors and responses and can adaptively learn complex functional forms, in particular, for situations where conventional regression models are ineffective. In a previous study, ANN with Bayesian regularization outperformed a benchmark linear model when predicting milk yield in dairy cattle or grain yield of wheat. Although breeding values rely on the assumption of additive inheritance, the predictive capabilities of ANN are of interest from the perspective of their potential to increase the accuracy of prediction of molecular breeding values used for genomic selection. This motivated the present study, in which the aim was to investigate the accuracy of ANN when predicting the expected progeny difference (EPD) of marbling score in Angus cattle. Various ANN architectures were explored, which involved two training algorithms, two types of activation functions, and from 1 to 4 neurons in hidden layers. For comparison, BayesCπ models were used to select a subset of optimal markers (referred to as feature selection), under the assumption of additive inheritance, and then the marker effects were estimated using BayesCπ with π set equal to zero. This procedure is referred to as BayesCpC and was implemented on a high-throughput computing cluster.

Results

The ANN with Bayesian regularization method performed equally well for prediction of EPD as BayesCpC, based on prediction accuracy and sum of squared errors. With the 3K-SNP panel, for example, prediction accuracy was 0.776 using BayesCpC, and ranged from 0.776 to 0.807 using BRANN. With the selected 700-SNP panel, prediction accuracy was 0.863 for BayesCpC and ranged from 0.842 to 0.858 for BRANN. However, prediction accuracy for the ANN with scaled conjugate gradient back-propagation was lower, ranging from 0.653 to 0.689 with the 3K-SNP panel, and from 0.743 to 0.793 with the selected 700-SNP panel.

Conclusions

ANN with Bayesian regularization performed as well as linear Bayesian regression models in predicting additive genetic values, supporting the idea that ANN are useful as universal approximators of functions of interest in breeding contexts.     

What does biologically meaningful mean? A perspective on gene regulatory network validation

Gene regulatory networks (GRNs) are rapidly being delineated, but their quality and biological meaning are often questioned. Here, I argue that biological meaning is challenging to define and discuss reasons why GRN validation should be interpreted cautiously.     

Quantification of Cre-mediated recombination by a novel strategy reveals a stable extra-chromosomal deletion-circle in mice

Background  

Inducible conditional knockout animals are widely used to get insight in the function of genes and the pathogenesis of human diseases. These models frequently rely on Cre-mediated recombination of sequences flanked by Lox-P sites. To understand the consequences of gene disruption, it is essential to know the efficiency of the recombination process.     

The anti-cyclic citrullinated peptide response in tuberculosis patients is not citrulline-dependent and sensitive to treatment

Introduction  

Patients with tuberculosis (TB) frequently produce anti-citrullinated protein antibodies (ACPA). The objective of this study is to characterize the citrulline-dependence of the ACPA reactivity in sera of patients with mycobacterium infections.     

Cross-adaptation and molecular modeling study of receptor mechanisms common to four taste stimuli in humans

Psychophysical cross-adaptation experiments were performed with two carbohydrates, sucrose (SUC) and fructose (FRU), and two sweeteners, acesulfame-K (MOD) and dulcin (DUL). Seven subjects were asked to match concentrations that elicited the same intensity as a sucrose reference (30 g/l). Cross-adaptation levels were calculated as the ratio of isointense concentrations measured for a given stimulus before and under adaptation. On average, cross-adaptation between SUC and FRU is low and apparently reciprocal. By contrast, cross-adaptation between SUC and MOD is clearly non-reciprocal: SUC adapts MOD significantly (24%, P < 0.005), but MOD fails to adapt SUC (2%, P < 0.79). Significant and reciprocal cross-enhancement is observed between DUL and MOD (approximately -20%, P < 0.03), and also between SUC and DUL (approximately -15%, P < 0.08). In parallel, molecular modeling of the four tastants was performed in order to look for the 12 common binding motifs that were isolated on 14 other tastants in a previous study. SUC and FRU each display 10 out of the 12 binding motifs, whereas DUL and MOD only display four and five distinct motifs respectively and do not have any motif in common. Experimental cross-adaptation levels seem to correlate well with the number of motifs that molecules have in common. FRU and SUC share a majority of binding motifs and correlatively show mutual cross-adaptation. Four motifs of MOD are found among the 10 motifs of SUC, which may explain why SUC cross-adapts MOD but not vice versa. By contrast, DUL and MOD do not share any motif and do not cross- adapt. The various molecular mechanisms that may be responsible for cross-adaptation and/or cross-enhancement are discussed in light of our results.      

Prediction and course of symptoms and lung function around an exacerbation in chronic obstructive pulmonary disease

Background

Frequent exacerbations induce a high burden to Chronic Obstructive Pulmonary Disease (COPD). We investigated the course of exacerbations in the published COSMIC study that investigated the effects of 1-year withdrawal of fluticasone after a 3-month run-in treatment period with salmeterol/fluticasone in patients with COPD.

Methods

In 373 patients, we evaluated diary cards for symptoms, Peak Expiratory Flow (PEF), and salbutamol use and assessed their course during exacerbations.

Results

There were 492 exacerbations in 224 patients. The level of symptoms of cough, sputum, dyspnea and nocturnal awakening steadily increased from 2 weeks prior to exacerbation, with a sharp rise during the last week. Symptoms of cough, sputum, and dyspnea reverted to baseline values at different rates (after 4, 4, and 7 weeks respectively), whereas symptoms of nocturnal awakening were still increased after eight weeks. The course of symptoms was similar around a first and second exacerbation. Increases in symptoms and salbutamol use and decreases in PEF were associated with a higher risk to develop an exacerbation, but with moderate predictive values, the areas under the receiver operating curves ranging from 0.63 to 0.70.

Conclusions

Exacerbations of COPD are associated with increased symptoms that persist for weeks and the course is very similar between a first and second exacerbation. COPD exacerbations are preceded by increased symptoms and salbutamol use and lower PEF, yet predictive values are too low to warrant daily use in clinical practice.     

Accurate prediction of protein secondary structure and solvent accessibility by consensus combiners of sequence and structure information

Background  

Structural properties of proteins such as secondary structure and solvent accessibility contribute to three-dimensional structure prediction, not only in the ab initio case but also when homology information to known structures is available. Structural properties are also routinely used in protein analysis even when homology is available, largely because homology modelling is lower throughput than, say, secondary structure prediction. Nonetheless, predictors of secondary structure and solvent accessibility are virtually always ab initio.