[신간의 별자리x] 우리/미술, 그리고 ‘슬픔의 박물관’

Finite mixture models are being used increasingly to model a wide variety of random phenomena for clustering, classification and density estimation. mclust is a powerful and popular package which allows modelling of data as a Gaussian finite mixture with different covariance structures and different numbers of mixture components, for a variety of purposes of analysis. Recently, version 5 of the package has been made available on CRAN. This updated version adds new covariance structures, dimension reduction capabilities for visualisation, model selection criteria, initialisation strategies for the EM algorithm, and bootstrap-based inference, making it a full-featured R package for data analysis via finite mixture modelling.

https://researchrepository.ucd.ie/bitstream/10197/7937/1/insight_publication.pdf

mclust 5: Clustering, Classification and Density Estimation Using Gaussian Finite Mixture Models.

Applied Discriminant Analysis

A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework.

Ever since Gregor Mendel deduced the probabilistic “laws” of heredity, mathematical and statistical methods have proved to be essential for genetic analyses. To keep pace both with the flood of genetic information pouring into journals and public databases and with our evolving knowledge of the complexity of genetic mechanisms, new models of genetic analysis are continually needed, particularly for traits measured in observational studies of humans. Mathematical and Statistical Methods for Genetic Analysis provides a solid bridge between modern genetics and advanced statistical methods, and it is filled with mathematical and statistical insights into the development of new genetic models. This book can be used not only as an advanced textbook for graduate students in statistics and mathematical modeling but also as a superb synthesis of recent statistical-genetics developments that any practitioner in this field will find immensely useful.As implied by the title and explicitly stated in the preface, this book is not intended as a cookbook for performing a genetic study; rather, the anticipated audience of this book is students who are already sophisticated in theoretical statistics, as well as in calculus and linear algebra. Readers with a suitable background are guided through fundamental statistical and mathematical concepts and then through a keen demonstration of how these methods can be adapted for genetic research.The book is composed of 13 chapters, of which the first describes the basic principles of population genetics; a brief appendix gives a summary of molecular genetics. Although these are nice summaries, I suspect that they would be too brief for mathematicians who have no background in biology; serious students of statistical genetics should look elsewhere for a fuller introduction to basic genetics. Chapter 2 covers the EM algorithm, with applications to ascertainment correction for segregation analysis. Chapter 3 discusses Newton's Method and Scoring, which are indispensable methods for obtaining maximum-likelihood estimates, and it illustrates their application for empirical Bayes estimation of allele frequencies. Some topics related to categorical data analysis, such as the transmission/disequilibrium test and tests for clustering of multinomial data, are covered in chapter 4. Kinship coefficients and their generalizations to more than two people are clearly presented in chapter 5, along with efficient recursive computational methods. These ideas of probabilistic descriptions of genetic relationships are key to genetic applications and are further expanded in chapter 6—for example, by partitioning the covariances of a pedigree into their genetic contributions.Chapter 7 covers computational methods for pedigree analyses, a topic on which Lange has published extensively. Although this chapter is full of key ideas that will prove useful for both segregation and linkage analyses, it was disappointing to not see more discussion of the hidden Markov methods that are the basis of the Lander-and-Green algorithm used in popular genetic-analysis software. Nonetheless, the breadth of topics covered is impressive, and generous insights appear throughout. Chapters 8–13 cover the polygenic model (with novel approximations), Markov-Chain Monte Carlo methods, evolutionary trees, radiation-hybrid mapping, models of recombination, and Poisson approximation. The explanations offered for the Markov-Chain Monte Carlo approach and for Poisson approximation, two highly active areas of research, are self-contained and worth reading for any statistician interested in these topics. The general steps needed to derive a genetic model are provided in each chapter, but readers who wish to get the most out of this book should plan to sit down with pencil and paper to fill in the details. Each chapter includes 10–12 problems to work through that reinforce the mathematical concepts.Mathematical and Statistical Methods for Genetic Analysis synthesizes many of the key statistical and mathematical topics that are used for genetic analyses. Lange indicates that this book represents his polished notes from years of teaching at UCLA and the University of Michigan, as well as a synthesis of his own published work. This bias toward Lange's own work is perhaps the book's greatest strength, because it allows him to offer deeper statistical and mathematical explanations for genetic analyses than are found in most other sources. This book is a must-read for students wishing to move into the field of statistical genetics, as well as for practicing statisticians who wish to adapt familiar statistical methods for genetic analyses.

Mathematical and Statistical Methods for Genetic Analysis

Reliable age-at-death estimates from the adult skeleton are of fundamental importance in forensic anthropology, because it contributes to the identity parameters used in a medicolegal death investigation. However, reliable estimates are difficult because many traditional aging methods depend on a set of population-specific criteria derived from individuals of European and African descent. The absence of information on the potential differences in the aging patterns of underrepresented, especially Latinx, populations may hinder our efforts to produce useful age-at-death estimates. In response to these concerns, this study explores the utility of currently available aging techniques and whether population-specific aging methods among Latinx groups are needed. The authors obtained data from two skeletal collections representing modern individuals of Mexican and Puerto Rican origin. They examined five newly developed computational shape-based techniques using 3D laser scans of the pubic symphysis and one traditional bone-to-phase technique. A validation test of all computational and traditional methods was implemented, and new population-specific equations using the computational algorithms were generated and tested against a subsample. Results suggest that traditional and computational aging techniques applied to the pubic symphysis perform best with individuals within 35-45 years of age. Levels of bias and inaccuracy increase as chronological age increases, with overestimation of individuals younger than 35 years and underestimation of individuals older than 45 years. New regression models provided error rates comparable to, and in some occasions outperformed, the original computational models developed on white American males, but age estimates did not significantly improve. This study shows that population-specific models do not necessarily improve age estimates in Latinx samples. Results do suggest that computational methods can ultimately outperform the Suchey-Brooks method and provide improved objectivity when estimating age at death in Latinx samples.

/pdf/age-at-death-estimation-for-modern-populations-in-mexico-and-m7axju5ifc.pdf

Age-at-Death Estimation for Modern Populations in Mexico and Puerto Rico through the Use of 3D Laser Scans of the Pubic Symphysis

What are the effects that genetics has had on Anthropological research and how can we think anthropologically about Genetics? Just as genetic data have encouraged new hypotheses about human phenotypic variation, evolutionary history, population interaction, and environmental effects, so too has Anthropology offered to genetic studies a new interpretive locus in its history and perspective. This introduction examines how the fields of Anthropology and Genetics have arrived at a crucial moment at which their interaction requires careful examination and critical reflection. The papers discussed here exemplify how we may engage in such a trans-disciplinary conversation. They speak to the future of thoughtful interaction between genetic and anthropological literature and seek a new integration that embodies the holism of the human biological sciences.

Better together: Thinking anthropologically about genetics

2Department of Scientifijic Computing, Florida State University, Tallahassee, Florida, USA. 3Department of Anthropology, University of Illinois at Urbana–Champaign, Champaign, Illinois, USA. *Correspondence to: Jieun Kim, Department of Scientifijic Computing, Florida State University, 400 Dirac Science Library, Tallahassee, FL 32306-4120 USA. E-mail: jkim17@fsu.edu.

/pdf/thinking-computationally-about-forensics-anthropological-1kbfs7xoe0.pdf

Thinking Computationally about Forensics: Anthropological Perspectives on Advancements in Technologies, Data, and Algorithms

Assigning correct population affinity to a skeleton can contribute important information to an investigation—yet recent work highlights high error rates when classifying Latinos with a traditional tool, Fordisc 3.1 (FD3). Our study examines whether misclassification trends exist, and whether these can be used to infer population affinity. We examine the relationships among ancestry, geography, and FD3 misclassifications of Latinos using canonical variate analysis and unsupervised model‐based clustering of craniometrics. Northern Mexicans appear more strongly associated with FD3 references samples with elevated amounts of European ancestry (e.g., American Blacks and Whites), while Southern Mexicans are more strongly associated with FD3 reference samples with reduced amounts of European ancestry (e.g., Guatemalans and Native Americans). FD3 classifications revealed that Latinos exhibited lower posterior probabilities when compared to other common case demographics (Whites and African Americas), even when the classification was “correct.” We make recommendations for practitioner interpretation of FD3 reports for casework.

Understanding (mis)classification trends of Hispanics in Fordisc 3.1: Incorporating cranial morphology, microgeographic origin, and admixture proportions for interpretation

Bridget F. B. Algee-Hewitt

Papers

Age-at-Death Estimation for Modern Populations in Mexico and Puerto Rico through the Use of 3D Laser Scans of the Pubic Symphysis

Better together: Thinking anthropologically about genetics

Thinking Computationally about Forensics: Anthropological Perspectives on Advancements in Technologies, Data, and Algorithms

Understanding (mis)classification trends of Hispanics in Fordisc 3.1: Incorporating cranial morphology, microgeographic origin, and admixture proportions for interpretation

Comments on "Estimation of stature from cranial sutures in a South Indian male population" by P. P. J. Rao et al