Explore the vast range of titles available.

Understanding the relationship between amino acid sequence and protein function is a long-standing challenge with far-reaching scientific and translational implications. State-of-the-art alignment-based techniques cannot predict function for one-third of microbial protein sequences, hampering our ability to exploit data from diverse organisms. Here, we train deep learning models to accurately predict functional annotations for unaligned amino acid sequences across rigorous benchmark assessments built from the 17,929 families of the protein families database Pfam. The models infer known patterns of evolutionary substitutions and learn representations that accurately cluster sequences from unseen families. Combining deep models with existing methods significantly improves remote homology detection, suggesting that the deep models learn complementary information. This approach extends the coverage of Pfam by >9.5%, exceeding additions made over the last decade, and predicts function for 360 human reference proteome proteins with no previous Pfam annotation. These results suggest that deep learning models will be a core component of future protein annotation tools. A deep learning model predicts protein functional annotations for unaligned amino acid sequences.

Objective: To assess the long-term impact of undernutrition during specific periods of fetal life, upon central adiposity, control of feeding behaviour and locomotor activity. Design: Pregnant rats were fed a control or low-protein (LP) diet, targeted to early (LPE), mid (LPM) or late (LPL) pregnancy or throughout gestation (LPA). The offspring were studied at 9 and 18 months of age. Measurements: Adiposity was assessed by measuring weight of abdominal fat depots relative to body weight. Locomotor activity was assessed using an infrared sensor array system in both light and dark conditions. Hypothalamic expression of mRNA for galanin and the galanin 2 receptor (Gal2R) was determined using real-time PCR. Results: At 9 months, male rats exposed to LP in utero had less fat in the gonadal depot, but were of similar body weight to controls. By 18 months, the males of groups LPA and LPM had more abdominal and less subcutaneous fat. Females deposited more fat centrally than males between 9 and 18 months of age, and this was more marked in groups LPA and LPL. Food intake was greater in LPM males. Among females hypophagia was noted in groups LPA and LPL. Expression of galanin and Gal2R were unaffected by maternal diet. Total locomotor activity was reduced in LPE males and all LP females in the light but not in the dark. Conclusion: Locomotor activity and feeding behaviour in aged rats are subject to prenatal programming influences. Fetal undernutrition does not programme obesity in rats without postnatal dietary challenge.

Periodontal disease is a common chronic adult condition. The bacterium Porphyromonas gingivalis has been implicated in the aetiology of this disease, which causes destruction of the connective tissue and bone around the root area of the tooth. It has been observed that invading P. gingivalis bacteria trigger the release of cytokines such as interleukin 8 and tumour necrosis factor α, leading to elevated numbers and activity of polymorphonucleocytes (PMN). As a result of stimulation by bacterial antigens, PMN produce the reactive oxygen species (ROS) superoxide via the respiratory burst as part of the host response to infection. Patients with periodontal disease display increased PMN number and activity. It has been suggested that this proliferation results in a high degree of ROS release, culminating in heightened oxidative damage to gingival tissue, periodontal ligament and alveolar bone. Antioxidant constituents in plasma have been well-documented, being chiefly ascorbate, albumin and urate, and these are known to display sensitivity to dietary antioxidant intakes. The concentration of antioxidants in saliva does not appear to mirror those of plasma. The extent of dietary influence upon salivary antioxidant status is unclear. Urate is the predominant salivary antioxidant, with albumin and ascorbate providing minor contributions. Previous research has found reduced salivary antioxidant activity in patients suffering from periodontal disease. An improved understanding of the role antioxidants play in periodontitis, and the influence of nutrition on antioxidant status, may lead to a possible nutritional strategy for the treatment of periodontal disease.

Nephrogenesis in the rat starts mid-gestation and continues into lactation. Maternal low protein (LP) intake leads to renal injury in rats and associates with mild renal injury in humans. We hypothesized that LP during early nephrogenesis or throughout gestation would induce more renal injury in rat offspring than when LP was only present before nephrogenesis. Pregnant rats were fed LP diet (9% casein) at early gestation (LPE, day 0–7), mid (LPM, day 8–14), late (LPL, day 15–22) or throughout gestation (LPA, day 0–22) and compared to controls on 18% casein diet. Offspring were studied at 18 months. Renal injury was assessed by 24 h proteinuria, plasma urea, antioxidant enzyme activities, and apoptosis (Bax/Bcl2). Proteinuria was higher in LPM males and LPE and LPM females. In LPM males glutathione peroxidase activity was lower, while in LPE males catalase activity was higher. Antioxidants were not much affected in females. Bax expression was higher in LPM males and females, while Bcl2 expression was higher in LPA females. Thus even before nephrogenesis (day 0–7), LP impacted on renal integrity in adult life, while LP during a later phase (day 15–22) or throughout gestation had less effect. In summary, for aging rat kidney LP poses the greatest threat when restricted to early nephrogenesis.

This study assessed the impact of reduced dietary protein during specific periods of fetal life upon muscle fibre development in young rats. Pregnant rats were fed a control or low-protein (LP) diet at early (days 0–7 gestation, LPEarly), mid (days 8–14, LPMid), late (days 15–22, LPLate) or throughout gestation (days 0–22, LPAll). The muscle fibre number and composition in soleus and gastrocnemius muscles of the offspring were studied at 4 weeks of age. In the soleus muscle, both the total number and density of fast fibres were reduced in LPMid females (P = 0·004 for both, Diet × Sex × Fibre type interactions), while both the total number and density of glycolytic (non-oxidative) fibres were reduced in LPEarly, LPMid and LPLate (but not LPAll) offspring compared with controls (P < 0·001 for both, Diet × Fibre type interaction). In the gastrocnemius muscle, only the density of oxidative fibres was reduced in LPMid compared with control offspring (P = 0·019, Diet × Fibre type interaction), with the density of slow fibres being increased in LPAll males compared with control (P = 0·024, Diet × Sex × Fibre type interaction). There were little or no effects of maternal diet on fibre type diameters in the two muscles. In conclusion, a maternal low-protein diet mainly during mid-pregnancy reduced muscle fibre number and density in 4-week-old rats, but there were muscle-specific differences in the fibre types affected.

Understanding the relationship between amino acid sequence and protein function is a long-standing problem in molecular biology with far-reaching scientific implications. Despite six decades of progress, state-of-the-art techniques cannot annotate 1/3 of microbial protein sequences, hampering our ability to exploit sequences collected from diverse organisms. In this paper, we explore an alternative methodology based on deep learning that learns the relationship between unaligned amino acid sequences and their functional annotations across all 17929 families of the Pfam database. Using the Pfam seed sequences we establish rigorous benchmark assessments that use both random and clustered data splits to control for potentially confounding sequence similarities between train and test sequences. Using Pfam full, we report convolutional networks that are significantly more accurate and computationally efficient than BLASTp, while learning sequence features such as structural disorder and transmembrane helices. Our model co-locates sequences from unseen families in embedding space, allowing sequences from novel families to be accurately annotated. These results suggest deep learning models will be a core component of future protein function prediction tools. Footnotes * Updates: - Clustered split. - Clan level analysis. - Additional explanation/graphs about stratification by sequence identity. - Computational performance: scripts for reproducibility. Performance by single CPU core. - Additional analysis for embedding space and few-shot learning section. * https://www.kaggle.com/googleai/pfam-seed-random-split * https://console.cloud.google.com/storage/browser/brain-genomics-public/research/proteins/pfam/random_split * https://console.cloud.google.com/storage/browser/brain-genomics-public/research/proteins/pfam/clustered_split

The low cost of digital communication has given rise to the problem of email spam, which is unwanted, harmful, or abusive electronic content. In this thesis, we present several advances in the application of online machine learning methods for automatically filtering spam. We detail a sliding-window variant of Support Vector Machines that yields state of the art results for the standard online filtering task. We explore a variety of feature representations for spam data. We reduce human labeling cost through the use of efficient online active learning variants. We give practical solutions to the one-sided feedback scenario, in which users only give labeling feedback on messages predicted to be non-spam. We investigate the impact of class label noise on machine learning-based spam filters, showing that previous benchmark evaluations rewarded filters prone to overfitting in real-world settings and proposing several modifications for combating these negative effects. Finally, we investigate the performance of these filtering methods on the more challenging task of abuse filtering in blog comments. Together, these contributions enable more accurate spam filters to be deployed in real-world settings, with greater robustness to noise, lower computation cost and lower human labeling cost.

When confronted with a substance of unknown identity, researchers often perform mass spectrometry on the sample and compare the observed spectrum to a library of previously-collected spectra to identify the molecule. While popular, this approach will fail to identify molecules that are not in the existing library. In response, we propose to improve the library's coverage by augmenting it with synthetic spectra that are predicted using machine learning. We contribute a lightweight neural network model that quickly predicts mass spectra for small molecules. Achieving high accuracy predictions requires a novel neural network architecture that is designed to capture typical fragmentation patterns from electron ionization. We analyze the effects of our modeling innovations on library matching performance and compare our models to prior machine learning-based work on spectrum prediction.