Biology subjects

Surendra, A.

Publications and source records attributed to Surendra, A..

7 recordsLinked to original sources

bioRxiv2024

Projection Statistics ProST Online statistical assessment of group separation in data projection analysis

MotivationUnsupervised data projection for the determination of trends in the data, visualization of multidimensional data in a reduced dimension space or feature space reduction through combination of data is a major step in data mining. Methods such as Principal Component Analysis or t-Distribution Stochastic Neighbor Embedding are regularly used as one of the first steps in computational biology or omics investigation. However, the significance of the separation of sample groups by these methods generally relies on visual assessment. User-friendly application for different projection methods, each focusing on distinct data properties, are needed as well as a rigorous method for statistical determination of the significance of separation of groups of interest in each dataset. ResultsWe present Projection STatistics (ProST), a user-friendly solution for data projection analysis providing three unsupervised (PCA, t-SNE and UMAP) and one supervised (LDA) approach. For each method we are including a novel statistical investigation of the significance of group separation with Mann-Whitney U-rank or t-test analysis as well as necessary preprocessing steps. ProST provides an unbiased, objective application of the determination of the significance of the separation of measurement groups through either linear or manifold projection analysis with methods ranging from a focus on the separation of points based on major variances or on point proximities based on distance. AvailabilityThe ProST software application is freely available at https://complimet.ca/shiny/ProST/ with source code provided on https://github.com/complimet/prost. Contactdanny.salem@nrc-cnrc.gc.ca or Miroslava.cuperlovic-culf@nrc-cnrc.gc.ca Supplementary informationSupplementary help pages are provided at https://complimet.ca/shiny/ProST/.

bioinformatics↗

bioRxiv2024

Taking stock of selective logging in the Andaman Islands, India: recent & legacy effects of timber extraction, assisted natural regeneration and a revamped working plan

Forest management is an evolving balance between biodiversity conservation and economic needs. A shift in Andaman Islands Working Plan mandate in 2000s reflects this evolution. Our study independently assesses the impact of said policy change on post-logging recovery of forests in Baratang and Middle Andaman. In 2017-18, we placed seventy-six 0.49ha plots across evergreen and deciduous patches and compared large-tree ([≥]180cm girth) density and diversity across forests that were logged after 2005 focussing on sustainability, logged in 1990s focussing on timber, logged twice in 1990s and after 2005, and unlogged forests. We assessed forest regeneration in thirty 0.01ha plots along a coupe road within forests logged after 2005. Forests logged after 2005 had similar density of large trees as forests logged in 1990s (despite having 1/3rd the recovery time), indicating reduced offtake or better recruitment. Along the unlogged--once-logged--twice-logged gradient, the dominance of Pterocarpus dalbergioides in deciduous patches decreased while the dominance of Diptercarpus sp. in evergreen patches increased. Compared to natural regeneration, proportionately more deciduous saplings were planted in both evergreen and deciduous patches. The new working plan maintains timber stock but not diversity. We make six simple recommendations to better align practice with the Working Plan mandate. SynthesisPost-2005 timber extraction policy in the Andaman Islands is partially successful but long-term forest health, in line with the working plan mandate, requires (1) lower timber offtake from deciduous patches and (2) targeted assisted regeneration of non-timber tree species.

ecology↗

bioRxiv2024

Imputation for Lipidomics and Metabolomics (ImpLiMet): Online optimization and method selection for missing data imputation

MotivationMissing values are often unavoidable in modern high-throughput measurements due to various experimental or analytical reasons. Imputation, the process of replacing missing values in a dataset with estimated values, plays an important role in multivariate and machine learning analyses. Three missingness patterns have been conceptualized: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). Each describes unique dependencies between the missing and observed data. As a result, the optimal imputation method for each dataset depends on the type of data, the cause of the missing data, and the nature of relationships between the missing and observed data. The challenge is to identify the optimal imputation solution for a given dataset. ResultsImpLiMet: is a user-friendly UI-platform that enables users to impute missing data using eight different methods. For the users dataset, ImpLiMet can suggest the optimal imputation solution through a grid search-based investigation of the error rate for imputation across three missingness data simulations. The effect of imputation can be visually assessed by histogram, kurtosis and skewness analyses, as well as principal component analysis (PCA) comparing the impact of the chosen imputation method on the distribution and overall behaviour of the data. Availability and implementationImpLiMet is freely available at https://complimet.ca/shiny/implimet/ with software accessible at https://github.com/complimet/ImpLiMet Contactsteffanyann.bennett@uottawa.ca and miroslava.cuperlovic-culf@nrc-cnrc.gca. Supplementary informationSupplementary data are available at Bioinformatics Advances online.

bioinformatics↗

bioRxiv2024

Harmonising distributed tree inventory datasets across India can fill critical gaps in tropical ecology

AO_SCPLOWBSTRACTC_SCPLOWO_LIGlobal analyses of tree diversity and function are strongly biased geographically, with poor representation from forests of the Indian subcontinent. Even though data from India - representing two-thirds of the subcontinent and spanning a wide range of tree-based biomes - exists, a barrier to syntheses is the absence of accessible and standardised data. Further, with increasing human footprint across ecosystems, data from Indian landscapes, with their long history of human-nature interactions is a key link to understand the future of tropical forested landscapes. Given the long history of human-nature interactions and high human footprint in the region, accessible and standardized data from the subcontinent can enable understanding the future of tropical landscapes under increasing human footprints globally. C_LIO_LICombining literature searches with manual data retrieval, we assembled the INdia Tree Inventory dataset, INvenTree. INvenTree is the largest meta-dataset of peer-reviewed publications (n = 465) from 1991-2023 on geolocated plot-based tree inventories of multispecies communities from Indian ecosystems, in aggregate covering 4653.64 ha and all of its woody biomes. C_LIO_LIUsing INvenTree, we show extensive sampling across tropical moist and dry forests, the dominant ecosystem types in the country. However, most studies have low sampling effort (median sampled area = 2 ha) and data across studies is not openly accessible (73.33 % of studies representing 83.43% of the sampled area), potentially hindering inclusion into regional or global syntheses. C_LIO_LISignificantly, we show majority authorship from within the country; 82.8% of corresponding authors were from India and 73.33% of the studies had all authors affiliated with Indian institutions. We also identify ecological and conservation sampling priority regions based on forest cover and forest loss and set a blueprint for future sampling efforts in the country. C_LIO_LIBased on extensive Indian scholarship in forest ecology showcased through the INvenTree dataset, we see opportunity for regional collaboration to create scientific inferences that are larger and scalable, while prioritising data and knowledge equity. Harmonising these existing datasets, and synthesising historic and grey literature, will contribute enormously to understanding the human dimension of tropical ecology as well as informing regional management and conservation. C_LI

ecology↗

bioRxiv2022

Forest bird decline and community change over 19 years in long-isolated South Asian tropical rainforest fragments

Recent evidence of forest bird declines worldwide is attributed to climate change and its interactive effects with recent land-use changes such as forest loss and fragmentation, and avian life-history traits. In Asian tropical forests, such effects are poorly understood as long-term data are lacking from fragments that are long-isolated rather than recently fragmented. Here, we use data from ~2000 point-counts from bird surveys carried out between 2000 - 2005 and 2019 in 19 long-isolated (~80 y) South Asian tropical rainforest fragments to examine changes in bird species richness, density, and composition in relation to fragment area (0.7 - 4310 ha), habitat structure, and time. Over the 19 y timespan, despite stable fragment areas, we uncovered a 29% decline in rainforest bird density and 7% decline in individual-rarefied species richness of rainforest birds, while density and richness of open-country birds remained stable. With increasing fragment area, rainforest bird species richness (jackknife estimate) increased, while open country bird richness (individual-rarefied) and density decreased. Larger fragments housed more compositionally stable bird communities, while poorer habitat was associated with lower diversity of rainforest birds but higher diversity, density, and compositional variation of open-country birds. Threshold analysis however indicated relatively small area thresholds (~20 ha) for rainforest bird species abundance. Besides identifying alarming declines in rainforest birds, the study confirms some but not all predictions for bird diversity in long-isolated forest fragments with stable forest-matrix boundaries, indicating that small fragments and habitat quality also matter.

ecology↗

bioRxiv2022

METAbolomics data Balancing with Over-sampling Al-gorithms (META-BOA): an online resource for addressing class imbalance

MotivationClass imbalance, or unequal sample sizes between classes, is an increasing concern in machine learning for metabolomic and lipidomic data mining, which can result in overfitting for the over-represented class. Numerous methods have been developed for handling class imbalance, but they are not readily accessible to users with limited computational experience. Moreover, there is no resource that enables users to easily evaluate the effect of different over-sampling algorithms. ResultsMETAbolomics data Balancing with Over-sampling Algorithms (META-BOA) is a web-based application that enables users to select between four different methods for class balancing, followed by data visualization and classification of the sample to observe the augmentation effects. META-BOA outputs a newly balanced dataset, generating additional samples in the minority class, according to the users choice of Synthetic Minority Over-sampling Technique (SMOTE), Borderline-SMOTE (BSMOTE), Adaptive Synthetic (ADASYN), or Random Over-Sampling Examples (ROSE). META-BOA further displays both principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE) visualization of data pre- and post-over-sampling. Random forest classification is utilized to compare sample classification in both the original and balanced datasets, enabling users to select the most appropriate method for their analyses. Availability and implementationMETA-BOA is available at https://complimet.ca/meta-boa. Supplementary InformationSupplementary material is available at Bioinformatics online.

bioinformatics↗

bioRxiv2019

Differential expression feature extraction (DEFE) and its application in RNA-seq data analysis

In differential gene expression data analysis, one objective is to identify groups of co-expressed genes from a large dataset to detect the association between such a group of genes and a phenotypic trait. This is often done through a clustering approach, such as k-means or bipartition hierarchical clustering, based on particular similarity measures in the grouping process. In such a dataset, the gene differential expression itself is an innate attribute that can be used in the feature extraction process. For example, in a dataset consisting of multiple treatments versus their controls, the expression of a gene in each treatment would have three possible behaviors, up-, down- regulated, or unchanged. We propose here a differential expression feature extraction (DEFE) method by using a string consisting of three numerical values at each character to denote such behavior, i.e. 1=up, 2=down, and 0=unchanged, which results in up to 3B differential expression patterns across all B comparisons. This approach has been successfully applied in many datasets, of which we present in this study two sets of RNA-sequencing (RNA-seq) data on wheat challenged with stress related phytohormones or Fusarium graminearum, the causal agent of fusarium head blight (FHB), a devastating wheat disease to illustrate the algorithm. Combinations of multiple schemes of DEFE patterns revealed groups of genes putatively associated with resistance or susceptibility to FHB. DEFE enabled discovery of genes closely associated with defense related signaling molecules such as JAZ10, shikimate and chorismate biosynthesis pathway and groups of wheat genes with differential effects between more or less virulent strains of Fusarium graminearum.

genomics↗