Comparative species abundance modeling of Capitellidae (Annelida) in Tampa Bay, Florida, USA

Capitellid polychaetes are ubiquitous throughout the world’s oceans and are often encountered in high abundance. We used an extensive dataset of species abundance and distribution records of the Capitella capitata complex, C. aciculata, C. jonesi, Heteromastus filiformis, Mediomastus ambiseta, and M. californiensis from Tampa Bay, Florida, USA, as a model system of closely related species filling a similar ecological niche. We sought to (1) characterize the spatial distribution of each species, (2) determine if a single species abundance modeling strategy could be applied to them all, and (3) assess environmental drivers of species distribution and abundance. We found that all species had a zero-inflated abundance distribution and there was spatial autocorrelation by bay regions. Lorenz curves were an effective tool to assess spatial patterns of species abundance across large areas. Bay segment, depth, and dissolved oxygen were the most important environmental drivers. Modeling was accomplished by comparing 6 different approaches: 4 generalized additive models (GAMs: Poisson, negative binomial, Tweedie, and zero-inflated Poisson distributions), hurdle models, and boosted regression trees. There was no single model with top performance for every species. However, GAM-Tweedie and hurdle models performed well overall and may be useful for studies of other benthic marine invertebrates.

RNA-seq analyses of molecular abundance (RoMA) for detecting differential gene expression

MotivationVarious methods have been proposed, each with its own limitations. Some naive normal-based tests have low testing power with invalid normal distribution assumptions for RNA-seq read counts, whereas count-based methods lack a biologically meaningful interpretation and have limited capability for integration with other analysis packages for mRNA abundance. In this study, we propose an improved method, RoMA, to accurately detect differential expression and unlock the integration with upstream and downstream analyses on mRNA abundance in RNA-seq studies.ResultsRoMA incorporates information from both mRNA abundance and raw counts. Studies on simulated data and two real datasets showed that RoMA provides an accurate quantification of mRNA abundance and a data adjustment-tolerant DE analysis with high AUC, low FDR, and an efficient control of type I error rate. This study provides a valid strategy for mRNA abundance modeling and data analysis integration for RNA-seq studies, which will greatly facilitate the identification and interpretation of DE genes.Availability and implementationRoMA is available at https://github.com/GuoshuaiCai/[email protected] or [email protected]

The Study on Blind Unmixing for Hyperspectral Imagery

Hyperspectral is the frontiers of Remote Sensingdevelopment, which plays a more andmore important role in many fields. themixed pixels become an main obstacle to the in depth development forquantification imageryanalysis,This paper presented a novel approach based on independentcomponent analysis for hyperspectralunmixing,it introducing the constraints of abundance nonnegative and abundancesum-to-one，the purpose of our algorithm was not to findindependent components as decomposition results anymore．It developed an abundance modeling technique todescribe the statistical distribution of the data．Themodeling approach is capable of self-adaptation，andcan be applied to hyperspectral images with different characteristics．Experimental results demonstrated that the proposedapproach can obtain more accurate results than the other state-of-the-artapproaches．