Scalable Phylogeny Reconstruction with Disaggregated Near-memory Processing

Disaggregated computer architectures eliminate resource fragmentation in next-generation datacenters by enabling virtual machines to employ resources such as CPUs, memory, and accelerators that are physically located on different servers. While this paves the way for highly compute- and/or memory-intensive applications to potentially deploy all CPUs and/or memory resources in a datacenter, it poses a major challenge to the efficient deployment of hardware accelerators: input/output data can reside on different servers than the ones hosting accelerator resources, thereby requiring time- and energy-consuming remote data transfers that diminish the gains of hardware acceleration. Targeting a disaggregated datacenter architecture similar to the IBM dReDBox disaggregated datacenter prototype, the present work explores the potential of deploying custom acceleration units adjacently to the disaggregated-memory controller on memory bricks (in dReDBox terminology), which is implemented on FPGA technology, to reduce data movement and improve performance and energy efficiency when reconstructing large phylogenies (evolutionary relationships among organisms). A fundamental computational kernel is the Phylogenetic Likelihood Function (PLF), which dominates the total execution time (up to 95%) of widely used maximum-likelihood methods. Numerous efforts to boost PLF performance over the years focused on accelerating computation; since the PLF is a data-intensive, memory-bound operation, performance remains limited by data movement, and memory disaggregation only exacerbates the problem. We describe two near-memory processing models, one that addresses the problem of workload distribution to memory bricks, which is particularly tailored toward larger genomes (e.g., plants and mammals), and one that reduces overall memory requirements through memory-side data interpolation transparently to the application, thereby allowing the phylogeny size to scale to a larger number of organisms without requiring additional memory.

The complete chloroplast genome sequence of Quercus ningangensis and its phylogenetic implication

Quercus ningangensis is an economically and ecologically important tree species belonging to the family Fagaceae. In this study, the complete chloroplast (cp) genome of Q. ningangensis was sequenced and assembled, and 18 published cp genomes of Quercus were retrieved for genomic analyses (including sequence divergence, repeat elements, and structure) and phylogenetic inference. With this study, we found that complete cp genomes in Quercus are conserved, and we discovered a codon composition bias, which may be related to genomic content and genetic characteristics. In addition, we detected considerable structural variations in the expansion and contraction of inverted repeat regions. Six regions with relatively high variable (matK-rps16, psbC, ycf3 intron, rbcL, petA-psbJ, and ycf1) were detected by conducting a sliding window analysis, which has a high potential for developing effective genetic markers. Phylogenetic analysis based on Bayesian inference and maximum likelihood methods resulted in a robust phylogenetic tree of Quercus with high resolution for nearly all identified nodes. The phylogenetic relationships showed that the phylogenetic position of Q. ningangensis was located between Q. sichourensis and Q. acuta. The results of this study contribute to future research into the phylogenetic evolution of Quercus section Cyclobalanopsis (Fagaceae).

REMOTELY SENSED IMAGE FAST CLASSIFICATION AND SMART THEMATIC MAP PRODUCTION

Apps available for Smartphone, as well as software for GNSS/GIS devices, permit to easily mapping the localization and shape of an area by acquiring the vertices coordinates of its contour. This option is useful for remote sensing classification, supporting the detection of representative sample sites of a known cover type to use for algorithm training or to test classification results. This article aims to analyse the possibility to produce smart maps from remotely sensed image classification in rapid way: the attention is focalized on different methods that are compared to identify fast and accurate procedure for producing up-to-date and reliable maps. Landsat 8 OLI multispectral images of northern Sicily (Italy) are submitted to various classification algorithms to distinguish water, bare soil and vegetation. The resulting map is useful for many purposes: appropriately inserted in a larger database aimed at representing the situation in a space-time evolutionary scenario, it is suitable whenever you want to capture the variation induced in a scene, e.g. burnt areas identification, vegetated areas definition for tourist-recreational purposes, etc. Particularly, pixel-based classification approaches are preferred, and experiments are carried out using unsupervised (k-means), vegetation index (NDVI, Normalized Difference Vegetation Index), supervised (minimum distance, maximum likelihood) methods. Using test sites, confusion matrix is built for each method, and quality indices are calculated to compare the results. Experiments demonstrate that NDVI submitted to k-means algorithm allows the best performance for distinguishing not only vegetation areas but also water bodies and bare soils. The resulting thematic map is converted for web publishing.

OptAGAN: Entropy-based Finetuning on Text VAE-GAN

Transfer learning through large pre-trained models has changed the landscape of current applications in natural language processing (NLP). Recently Optimus, a variational autoencoder (VAE) which combines two pre-trained models, BERT and GPT-2, has been released, and its combination with generative adversarial networks (GANs) has been shown to produce novel, yet very human-looking text. The Optimus and GANs combination avoids the troublesome application of GANs to the discrete domain of text, and prevents the exposure bias of standard maximum likelihood methods. We combine the training of GANs in the latent space, with the finetuning of the decoder of Optimus for single word generation. This approach lets us model both the high-level features of the sentences, and the low-level word-by-word generation. We finetune using reinforcement learning (RL) by exploiting the structure of GPT-2 and by adding entropy-based intrinsically motivated rewards to balance between quality and diversity. We benchmark the results of the VAE-GAN model, and show the improvements brought by our RL finetuning on three widely used datasets for text generation, with results that greatly surpass the current state-of-the-art for the quality of the generated texts.

Assessing Causal Relationship Between Human Blood Metabolites and Five Neurodegenerative Diseases With GWAS Summary Statistics

Background: Neurodegenerative diseases (NDDs) are the leading cause of disability worldwide while their metabolic pathogenesis is unclear. Genome-wide association studies (GWASs) offer an unprecedented opportunity to untangle the relationship between metabolites and NDDs.Methods: By leveraging two-sample Mendelian randomization (MR) approaches and relying on GWASs summary statistics, we here explore the causal association between 486 metabolites and five NDDs including Alzheimer’s Disease (AD), amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), Parkinson’s disease (PD), and multiple sclerosis (MS). We validated our MR results with extensive sensitive analyses including MR-PRESSO and MR-Egger regression. We also performed linkage disequilibrium score regression (LDSC) and colocalization analyses to distinguish causal metabolite-NDD associations from genetic correlation and LD confounding of shared causal genetic variants. Finally, a metabolic pathway analysis was further conducted to identify potential metabolite pathways.Results: We detected 164 metabolites which were suggestively associated with the risk of NDDs. Particularly, 2-methoxyacetaminophen sulfate substantially affected ALS (OR = 0.971, 95%CIs: 0.961 ∼ 0.982, FDR = 1.04E-4) and FTD (OR = 0.924, 95%CIs: 0.885 ∼ 0.964, FDR = 0.048), and X-11529 (OR = 1.604, 95%CIs: 1.250 ∼ 2.059, FDR = 0.048) and X-13429 (OR = 2.284, 95%CIs: 1.457 ∼ 3.581, FDR = 0.048) significantly impacted FTD. These associations were further confirmed by the weighted median and maximum likelihood methods, with MR-PRESSO and the MR-Egger regression removing the possibility of pleiotropy. We also observed that ALS or FTD can alter the metabolite levels, including ALS and FTD on 2-methoxyacetaminophen sulfate. The LDSC and colocalization analyses showed that none of the identified associations could be driven by genetic correlation or confounding by LD with common causal loci. Multiple metabolic pathways were found to be involved in NDDs, such as “urea cycle” (P = 0.036), “arginine biosynthesis” (P = 0.004) on AD and “phenylalanine, tyrosine and tryptophan biosynthesis” (P = 0.046) on ALS.Conclusion: our study reveals robust bidirectional causal associations between servaral metabolites and neurodegenerative diseases, and provides a novel insight into metabolic mechanism for pathogenesis and therapeutic strategies of these diseases.

No longer endemic to Africa: Kribiodosis Kieffer, 1921 (Diptera, Chironomidae) new to Oriental China with a phylogeny and expanded adult generic diagnoses

Kribiodosis Kieffer, 1921, an African genus of Chironomini (Diptera: Chironomidae), is newly recorded from the Oriental region through a new species K. cantonensis sp. n. Detailed descriptions of the male, female and a DNA barcode are provided. With the inclusion of the new species bearing scutal tubercle and fused tibial comb, the generic diagnosis needs revision and expansion. The phylogenetic position of Kribiodosis within the tribe Chironomini is explored based on five concatenated genetic makers (18S, 28S, CAD1, CAD4 and COI-3P) using both mixed-model Bayesian inference and maximum likelihood methods. Kribiodosis is placed as a core member of the Microtendipes group but its precise sister group remains unclear. Inclusion of the analysis of Nilodosis Kieffer, another Chironomini genus with an African-Oriental distribution, reveals an unexpected robust position as sister to a large and diverse inclusive group of many Chironomini.

18S and ITS2 rDNA sequence-structure phylogeny of Prototheca (Chlorophyta, Trebouxiophyceae)

Protothecosis is an infectious disease caused by organisms currently classified within the green algal genus Prototheca. The disease can manifest as cutaneous lesions, olecranon bursitis or disseminated or systemic infections in both immunocompetent and immunosuppressed patients. Concerning diagnostics, taxonomic validity is important. Prototheca, closely related to the Chlorella species complex, is known to be polyphyletic, branching with Auxenochlorella and Helicosporidium. The phylogeny of Prototheca was discussed and revisited several times in the last decade; new species have been described. Phylogenetic analyses were performed using ribosomal DNA (rDNA) and partial mitochondrial cytochrome b (cytb) sequence data. In this work we use Internal Transcribed Spacer 2 (ITS2) as well as 18S rDNA data. However, for the first time, we reconstruct phylogenetic relationships of Prototheca using primary sequence and RNA secondary structure information simultaneously, a concept shown to increase robustness and accuracy of phylogenetic tree estimation. Using encoded sequence-structure data, Neighbor-Joining, Maximum-Parsimony and Maximum-Likelihood methods yielded well-supported trees in agreement with other trees calculated on rDNA; but differ in several aspects from trees using cytb as a phylogenetic marker. ITS2 secondary structures of Prototheca sequences are in agreement with the well-known common core structure of eukaryotes but show unusual differences in their helix lengths. An elongation of the fourth helix of some species seems to have occurred independently in the course of evolution.

Mitochondrial genome of the critically endangered Baer’s Pochard, Aythya baeri, and its phylogenetic relationship with other Anatidae species

Historically, the diving duck, Baer’s Pochard (Aythya baeri) was widely distributed in East and South Asia, but according to a recent estimate, its global population is now less than 1000 individuals. To date, the mitochondrial genome of A. baeri has not been deposited and is not available in GenBank. Therefore, we aimed to sequence the complete mitochondrial genome of this species. The genome was 16,623 bp in length, double stranded, circular in shape, and contained 13 protein-coding genes, 22 tRNA genes, two rRNA genes, and one non-coding control region. Many structural and compositional similarities were discovered between A. baeri and the other three Aythya mitochondrial genomes. Among 13 protein-coding genes of the four Aythya species, the fastest-evolving gene was ATP8 while the slowest-evolving gene was COII. Furthermore, the phylogenetic tree of Anatidae based on Bayesian inference and maximum likelihood methods showed that the relationships among 15 genera of the Anatidae family were as follows: Dendrocygna was an early diverging lineage that was fairly distant from the other ingroup taxa; Cygnus, Branta, and Anser were clustered into one branch that corresponded to the Anserinae subfamily; and Aythya, Asarcornis, Netta, Anas, Mareca, Mergus, Lophodytes, Bucephala, Tadorna, Cairina, and Aix were clustered into another branch that corresponded to the Anatinae subfamily. Our target species and three other Aythya species formed a monophyletic group. These results provide new mitogenomic information to support further phylogenetic and taxonomic studies and genetic conservation of Anatidae species.

PERBANDINGAN PRINCIPAL AXIS FACTORING DAN MAXIMUM LIKELIHOOD DALAM MENENTUKAN FAKTOR DOMINAN YANG MEMPENGARUHI PEMBELAJARAN NAHWU SHOROF (STUDI KASUS PONDOK PESANTREN ROUDLOTUL MUTA’ALLIMIN PUTRI)

This study discusses the dominant factors that influence the success of learning nahwu shorof at the Roudlotul Mutaalimin Islamic Boarding School for the daughter of Minggirsari Village. Determining the dominant factor is done to maximize the quality of education in the boarding school, so that the interest of prospective students is increasing. In this study, two extraction methods were compared, namely the Principal Axis Factoring and Maximum Likelihood methods. There are 13 variables that affect the success of nahwu shorof learning, namely the natural environment (P1), social environment (P23), curriculum (P49), madrasa program (P1012), facilities and facilities (P1315), teaching staff (P1619), condition of physiological (P2021), condition of the five senses (P22), interest in learning (P2325), intelligence of students (P26), student talent (P27), motivation of students (P28), cognitive ability (P2930). The purpose of this study, namely to determine the most appropriate extraction method used in the analysis. The result of this study is the Maximum Likelihood method which is more appropriate than the Principal Axis Factoring method, because it has a smaller RMSE (Root Mean Squared Error) value.