The tubercle complex consists of closely related mycobacterium species which appear to be variants of a single species. Comparative genome analysis of different strains could provide useful clues and insights into the genetic diversity of the species. We integrated genome assemblies of 96 strains from Mycobacterium tuberculosis complex (MTBC), which included 8 Indian clinical isolates sequenced and assembled in this study, to understand its pangenome architecture. We predicted genes for all the 96 strains and clustered their respective CDSs into homologous gene clusters (HGCs) to reveal a hard-core, soft-core and accessory genome component of MTBC. The hard-core (HGCs shared amongst 100% of the strains) was comprised of 2,066 gene clusters whereas the soft-core (HGCs shared amongst at least 95% of the strains) comprised of 3,374 gene clusters. The change in the core and accessory genome components when observed as a function of their size revealed that MTBC has an open pangenome. We identified 74 HGCs that were absent from reference strains H37Rv and H37Ra but were present in most of clinical isolates. We report PCR validation on 9 candidate genes depicting 7 genes completely absent from H37Rv and H37Ra whereas 2 genes shared partial homology with them accounting to probable insertion and deletion events. The pangenome approach is a promising tool for studying strain specific genetic differences occurring within species. We also suggest that since selecting appropriate target genes for typing purposes requires the expected target gene be present in all isolates being typed, therefore estimating the core-component of the species becomes a subject of prime importance.

Scalability and accuracy are well recognized challenges in deep extreme multi-label learning where the objective is to train architectures for automatically annotating a data point with the most relevant subset of labels from an extremely large label set. This paper develops the DeepXML framework that addresses these challenges by decomposing the deep extreme multi-label task into four simpler sub-tasks each of which can be trained accurately and efficiently. Choosing different components for the four sub-tasks allows Deep-XML to generate a family of algorithms with varying trade-offs between accuracy and scalability. In particular, DeepXML yields the Astec algorithm that could be 2-12% more accurate and 5-30× faster to train than leading deep extreme classifiers on publically available short text datasets. Astec could also efficiently train on Bing short text datasets containing up to 62 million labels while making predictions for billions of users and data points per day on commodity hardware. This allowed Astec to be deployed on the Bing search engine for a number of short text applications ranging from matching user queries to advertiser bid phrases to showing personalized ads where it yielded significant gains in click-through-rates, coverage, revenue and other online metrics over state-of-the-art techniques currently in production. DeepXML's code is available at https://github.com/Extreme-classification/deepxml.

No abstract

Optimizing the placement of the wind turbines in a wind farm to achieve optimal performance is an active area of research, with numerous research studies being published every year. Typically, the area available for the wind farm is divided into cells (a cell may/may not contain a wind turbine) and an optimization algorithm is used. In this study, the effect of the cell size on the optimal layout is being investigated by reducing it from five rotor diameter (previous studies) to 1/40 rotor diameter (present study). A code is developed for optimizing the placement of wind turbines in large wind farms. The objective is to minimize the cost per unit power produced from the wind farm. A genetic algorithm is employed for the optimization. The velocity deficits in the wake of the wind turbines are estimated using a simple wake model. The code is verified using the results from the previous studies. Results are obtained for three different wind regimes: (1) Constant wind speed and fixed wind direction, (2) constant wind speed and variable wind direction, and (3) variable wind speed and variable wind direction. Cost per unit power is reduced by 11.7% for Case 1, 11.8% for Case 2, and 15.9% for Case 3 for results obtained in this study. The advantages/benefits of a refined grid spacing of 1/40 rotor diameter (1 m) are evident and are discussed. To get an understanding of the sensitivity of the power produced to the wake model, optimized layout is obtained for the Case 1 using a different wake model.

Extreme multi-label classification (XML) involves tagging a data point with its most relevant subset of labels from an extremely large label set, with several applications such as product-to-product recommendation with millions of products. Although leading XML algorithms scale to millions of labels, they largely ignore label metadata such as textual descriptions of the labels. On the other hand, classical techniques that can utilize label metadata via representation learning using deep networks struggle in extreme settings. This paper develops the DECAF algorithm that addresses these challenges by learning models enriched by label metadata that jointly learn model parameters and feature representations using deep networks and offer accurate classification at the scale of millions of labels. DECAF makes specific contributions to model architecture design, initialization, and training, enabling it to offer up to 2-6% more accurate prediction than leading extreme classifiers on publicly available benchmark product-to-product recommendation datasets, such as LF-AmazonTitles-1.3M. At the same time, DECAF was found to be up to 22× faster at inference than leading deep extreme classifiers, which makes it suitable for real-time applications that require predictions within a few milliseconds. The code for DECAF is available at the following URL https://github.com/Extreme-classification/DECAF.