Search

Scientists building the Tree of Life face an overwhelming challenge to categorize phenotypes (e.g., anatomy, physiology) from millions of living and fossil species. This biodiversity challenge far outstrips the capacities of trained scientific experts. Here we explore whether crowdsourcing can be used to collect matrix data on a large scale with the participation of nonexpert students, or "citizen scientists." Crowdsourcing, or data collection by nonexperts, frequently via the internet, has enabled scientists to tackle some large-scale data collection challenges too massive for individuals or scientific teams alone. The quality of work by nonexpert crowds is, however, often questioned and little data have been collected on how such crowds perform on complex tasks such as phylogenetic character coding. We studied a crowd of over 600 nonexperts and found that they could use images to identify anatomical similarity (hypotheses of homology) with an average accuracy of 82% compared with scores provided by experts in the field. This performance pattern held across the Tree of Life, from protists to vertebrates. We introduce a procedure that predicts the difficulty of each character and that can be used to assign harder characters to experts and easier characters to a nonexpert crowd for scoring. We test this procedure in a controlled experiment comparing crowd scores to those of experts and show that crowds can produce matrices with over 90% of cells scored correctly while reducing the number of cells to be scored by experts by 50%. Preparation time, including image collection and processing, for a crowdsourcing experiment is significant, and does not currently save time of scientific experts overall. However, if innovations in automation or robotics can reduce such effort, then large-scale implementation of our method could greatly increase the collective scientific knowledge of species phenotypes for phylogenetic tree building. For the field of crowdsourcing, we provide a rare study with ground truth, or an experimental control that many studies lack, and contribute new methods on how to coordinate the work of experts and nonexperts. We show that there are important instances in which crowd consensus is not a good proxy for correctness., (© The Author(s) 2017. Published by Oxford University Press on behalf of the Systematic Biologists. All rights reserved. For Permissions, please email: journals.permissions@oup.com.)

The phenotype represents a critical interface between the genome and the environment in which organisms live and evolve. Phenotypic characters also are a rich source of biodiversity data for tree building, and they enable scientists to reconstruct the evolutionary history of organisms, including most fossil taxa, for which genetic data are unavailable. Therefore, phenotypic data are necessary for building a comprehensive Tree of Life. In contrast to recent advances in molecular sequencing, which has become faster and cheaper through recent technological advances, phenotypic data collection remains often prohibitively slow and expensive. The next-generation phenomics project is a collaborative, multidisciplinary effort to leverage advances in image analysis, crowdsourcing, and natural language processing to develop and implement novel approaches for discovering and scoring the phenome, the collection of phentotypic characters for a species. This research represents a new approach to data collection that has the potential to transform phylogenetics research and to enable rapid advances in constructing the Tree of Life. Our goal is to assemble large phenomic datasets built using new methods and to provide the public and scientific community with tools for phenomic data assembly that will enable rapid and automated study of phenotypes across the Tree of Life.

This article reviews five approximate statistical tests for determining whether one learning algorithm outperforms another on a particular learning task. These tests are compared experimentally to determine their probability of incorrectly detecting a difference when no difference exists (type I error). Two widely used statistical tests are shown to have high probability of type I error in certain situations and should never be used: a test for difference of two proportions and a paired-differences t test based on taking several random train-test splits. A third test, a paired-differences t test based on 10-fold cross-validation, exhibits somewhat elevated probability of type I error. A fourth test, McNemar's test, is shown to have low type I error. The fifth test is a new test, 5 x 2 cv, based on five iterations of twofold cross-validation. Experiments show that this test also has acceptable type I error. The article also measures the power (ability to detect algorithm differences when they do exist) of these tests. The cross-validated t test is the most powerful. The 5 x 2 cv test is shown to be slightly more powerful than McNemar's test. The choice of the best test is determined by the computational cost of running the learning algorithm. For algorithms that can be executed only once, McNemar's test is the only test with acceptable type I error. For algorithms that can be executed 10 times, the 5 x 2 cv test is recommended, because it is slightly more powerful and because it directly measures variation due to the choice of training set.

Building predictive models for iterative drug design in the absence of a known target protein structure is an important challenge. We present a novel technique, Compass, that removes a major obstacle to accurate prediction by automatically selecting conformations and alignments of molecules without the benefit of a characterized active site. The technique combines explicit representation of molecular shape with neural network learning methods to produce highly predictive models, even across chemically distinct classes of molecules. We apply the method to predicting human perception of musk odor and show how the resulting models can provide graphical guidance for chemical modifications.