Identification and utilization of genetic determinants of trait measurement errors in image-based,high-throughput phenotyping

The accuracy of trait measurements greatly affects the quality of genetic analyses. During automated phenotyping, trait measurement errors, i.e. differences between automatically extracted trait values and ground truth, are often treated as random effects that can be controlled by increasing population sizes and/or replication number. In contrast, there is some evidence that trait measurement errors may be partially under genetic control. Consistent with this hypothesis, we observed substantial nonrandom, genetic contributions to trait measurement errors for five maize (Zea mays) tassel traits collected using an image-based phenotyping platform. The phenotyping accuracy varied according to whether a tassel exhibited “open” versus. “closed” branching architecture, which is itself under genetic control. Trait-associated SNPs (TASs) identified via genome-wide association studies (GWASs) conducted on five tassel traits that had been phenotyped both manually (i.e. ground truth) and via feature extraction from images exhibit little overlap. Furthermore, identification of TASs from GWASs conducted on the differences between the two values indicated that a fraction of measurement error is under genetic control. Similar results were obtained in a sorghum (Sorghum bicolor) plant height dataset, demonstrating that trait measurement error is genetically determined in multiple species and traits. Trait measurement bias cannot be controlled by increasing population size and/or replication number.

The accuracy of high-throughput phenotyping can be affected by genetically determined measurement biases, which can alter the results of genetic analyses.     

Optimising experimental design for high-throughput phenotyping in mice: a case study

To further the functional annotation of the mammalian genome, the Sanger Mouse Genetics Programme aims to generate and characterise knockout mice in a high-throughput manner. Annually, approximately 200 lines of knockout mice will be characterised using a standardised battery of phenotyping tests covering key disease indications ranging from obesity to sensory acuity. From these findings secondary centres will select putative mutants of interest for more in-depth, confirmatory experiments. Optimising experimental design and data analysis is essential to maximise output using the resources with greatest efficiency, thereby attaining our biological objective of understanding the role of genes in normal development and disease. This study uses the example of the noninvasive blood pressure test to demonstrate how statistical investigation is important for generating meaningful, reliable results and assessing the design for the defined research objectives. The analysis adjusts for the multiple-testing problem by applying the false discovery rate, which controls the number of false calls within those highlighted as significant. A variance analysis finds that the variation between mice dominates this assay. These variance measures were used to examine the interplay between days, readings, and number of mice on power, the ability to detect change. If an experiment is underpowered, we cannot conclude whether failure to detect a biological difference arises from low power or lack of a distinct phenotype, hence the mice are subjected to testing without gain. Consequently, in confirmatory studies, a power analysis along with the 3Rs can provide justification to increase the number of mice used.     

A high-throughput method for the quantitative analysis of auxins

Auxin measurements in plants are critical to understanding both auxin signaling and metabolic homeostasis. The most abundant natural auxin is indole-3-acetic acid (IAA). This protocol is for the precise, high-throughput determination of free IAA in plant tissue by isotope dilution analysis using gas chromatography-mass spectrometry (GC-MS). The steps described are as follows: harvesting of plant material; amino and polymethylmethacrylate solid-phase purification followed by derivatization with diazomethane (either manual or robotic); GC-MS analysis; and data analysis. [13C?]IAA is the standard used. The amount of tissue required is relatively small (25 mg of fresh weight) and one can process more than 500 samples per week using an automated system. To extract eight samples, this procedure takes ～3 h, whether performed manually or robotically. For processing more than eight samples, robotic extraction becomes substantially more time efficient, saving at least 0.5 h per additional batch of eight samples.     

Exploring the elephant: histopathology in high-throughput phenotyping of mutant mice

Recent advances in gene knockout techniques and the in vivo analysis of mutant mice, together with the advent of large-scale projects for systematic mouse mutagenesis and genome-wide phenotyping, have allowed the creation of platforms for the most complete and systematic analysis of gene function ever undertaken in a vertebrate. The development of high-throughput phenotyping pipelines for these and other large-scale projects allows investigators to search and integrate large amounts of directly comparable phenotype data from many mutants, on a genomic scale, to help develop and test new hypotheses about the origins of disease and the normal functions of genes in the organism. Histopathology has a venerable history in the understanding of the pathobiology of human and animal disease, and presents complementary advantages and challenges to in vivo phenotyping. In this review, we present evidence for the unique contribution that histopathology can make to a large-scale phenotyping effort, using examples from past and current programmes at Lexicon Pharmaceuticals and The Jackson Laboratory, and critically assess the role of histopathology analysis in high-throughput phenotyping pipelines.     

A high-throughput screen for MscL channel activity and mutational phenotyping

A novel fluorescence-based screen for bacterial mechanosensitive ion-channel activity has been developed. This assay is capable of clearly distinguishing the previously observed gain of function and loss of function phenotypes for the Escherichia coli mechanosensitive channel of large conductance (Ec-MscL). The method modifies Molecular Probes' Live/Dead BacLight bacterial viability assay to monitor MscL channel activity as a function of bacterial survival from osmotic downshock.     

Principal component analysis of quantitative trait loci for immune response to adenovirus in mice

Data on the duration of transgene expression in the liver, the presence of cytotoxic T lymphocytes (CTLs) against adenovirus, and serum cytokines from 18 strains of C57BL/6 x DBA/2 (B x D) recombinant inbred mice were analyzed. Our aim was to detect quantitative trait loci (QTLs) that may have causal relationship with the duration of adenovirus-mediated transgene expression in the liver. Information from beta-galactosidase (LacZ) expression; CTL production; and serum levels of gamma interferon, tumor necrosis factor-alpha, and interleukin-6 30 days after intravenous injection of liver LacZ were summarized by principal component analysis and analyzed using maximum likelihood interval mapping implemented in the QTL cartographer software. Two principal component (PC) scores explained 82.5% of the phenotypic variance in the original variables and identified QTLs not identified by analysis of individual traits. The distribution of original variables among PCs was such that variables in PC1 were predominantly cytokines with little CTL response whereas LacZ and CTL were the predominant contributors to PC2 with practically no contribution from cytokines. PC1 was significantly associated with two QTLs on chromosomes 7 and 9 located at 57.5 cM and 41.01 cM, respectively. Five QTLs were significantly associated with PC2 on chromosomes 12 (23.01 and 31.01 cM) and 15 (29.21, 36.01, and 56.31 cM). These results illustrate the use of principal component analysis in mapping QTLs using multiple correlated traits.     

Analysis of quantitative trait loci for behavioral laterality in mice

Laterality is believed to have genetic components, as has been deduced from family studies in humans and responses to artificial selection in mice, but these genetic components are unknown and the underlying physiological mechanisms are still a subject of dispute. We measured direction of laterality (preferential use of left or right paws) and degree of laterality (absolute difference between the use of left and right paws) in C57BL/6ByJ (B) and NZB/BlNJ (N) mice and in their F(1) and F(2) intercrosses. Measurements were taken of both forepaws and hind paws. Quantitative trait loci (QTL) did not emerge for direction but did for degree of laterality. One QTL for forepaw (LOD score = 5.6) and the second QTL for hind paw (LOD score = 7.2) were both located on chromosome 4 and their peaks were within the same confidence interval. A QTL for plasma luteinizing hormone concentration was also found in the confidence interval of these two QTL. These results suggest that the physiological mechanisms underlying degree of laterality react to gonadal steroids.     

A novel assay for monoacylglycerol hydrolysis suitable for high-throughput screening

A simple assay for monoacylglycerol hydrolysis suitable for high-throughput screening is described. The assay uses [(3)H]2-oleoylglycerol as substrate, with the tritium label in the glycerol part of the molecule and the use of phenyl sepharose gel to separate the hydrolyzed product ([(3)H]glycerol) from substrate. Using cytosolic fractions derived from rat cerebella as a source of hydrolytic activity, the assay gives the appropriate pH profile and sensitivity to inhibition with compounds known to inhibit hydrolysis of this substrate. The assay could also be adapted to a 96-well plate format, using C6 cells as the source of hydrolytic activity. Thus the assay is simple and appropriate for high-throughput screening of inhibitors of monoacylglycerol hydrolysis.     

A sensitive, simple assay of mitochondrial ATP synthesis of cultured mammalian cells suitable for high-throughput analysis

A new assay has been developed to measure mitochondrial ATP synthesis of cultured mammalian cells. Cells in a microplate are exposed to streptolysin O to make plasma membranes permeable without damaging mitochondrial function and ATP synthesis is monitored by luciferase. Addition of inhibitors of F_oF₁-ATP synthase (F_oF₁), respiratory chain, TCA cycle and ATP/ADP translocator, as well as knockdown of β-subunit of F_oF₁, resulted in loss of ATP synthesis. Compared with the conventional procedures that need mitochondria fractionation and detergent, this assay is simple, sensitive and suitable for high-throughput analysis of genes and drugs that could affect mitochondrial functional integrity as represented by ATP synthesis activity.     

Measuring growth rate in high-throughput growth phenotyping

Growth rate is an important variable and parameter in biology with a central role in evolutionary, functional genomics, and systems biology studies. In this review the pros and cons of the different technologies presently available for high-throughput measurements of growth rate are discussed. Growth rate can be measured in liquid microcultivation of individual strains, in competition between strains, as growing colonies on agar, as division of individual cells, and estimated from molecular reporters. Irrespective of methodology, statistical issues such as spatial biases and batch effects are crucial to investigate and correct for to ensure low false discovery rates. The rather low correlations between studies indicate that cross-laboratory comparison and standardization are pressing issue to assure high-quality and comparable growth-rate data.     

Fine mapping of Ath6, a quantitative trait locus for atherosclerosis in mice

Ath6 is a novel quantitative trait locus associated with differences in susceptibility to atherosclerosis between C57BL/6J (B6) and C57BLKS/J (BKS) inbred mouse strains. Combining data from an intercross and a backcross (1593 meioses) between mice from B6 and BKS strains and from The Jackson Laboratory interspecific backcross panels, (C57BL/6J ×Mus spretus) F₁× C57BL/6J and (C57BL/6J × SPRET/Ei) F₁× SPRET/Ei, we constructed a consensus genetic map and narrowed Ath6 to a 1.07 ± 0.26 cM interval between the anonymous DNA marker D12Pgn4 and the gene Nmyc1. This region is near the proximal end of murine Chromosome (Chr) 12, which is homologous to the human chromosomal region 2p24-p25. Marker order in the Ath6 region was concordant among the two crosses and The Jackson Laboratory interspecific backcross panels. This high resolution map rules out candidate genes encoding apolipoprotein B, syndecan 1, and Adam17. The two Ath6 crosses have a combined potential resolution of 0.06 cM. Received: 12 September 2000 / Accepted: 22 February 2001     

Identification of quantitative trait loci for prolificacy and growth in mice

Marker–quantitative trait locus (QTL) linkage was evaluated in F₂ intercross and backcross mouse populations derived from stocks differing dramatically in prolificacy and mature weight. A highly prolific outbred Quackenbush-Swiss mouse line, or an inbred line derived from it (16.62 ± 0.22 and 14.64 ± 0.27 pups per litter, respectively) were used as one of the grandparents in these populations. The less prolific C57BL/6J inbred mouse line (6.67 ± 0.37 pups per litter) was used as the other grandparent. Linkage was evaluated in a three-step process that involved selective genotyping of F₂ intercross progeny representing extremes for prolificacy, genotyping of the full F₂ for chromosomal regions potentially associated with prolificacy, and genotyping of the backcross for genomic regions significantly associated with prolificacy in the F₂. Segments of Chromosomes (Chrs) 2 and 4 were significantly (P < 0.05, experiment-wise error rate) associated with prolificacy, and LOD scores suggestive of linkage were observed for litter size on Chr 9 and growth on Chrs 4 and 11. Existence of growth QTL was also supported by marker effects that were significant (P < 0.05) or approaching significance (P < 0.10) in the backcross. Additive litter size QTL effects ranged from 0.56 to 0.79 pups per litter, and dominance deviations ranged from −0.56 to 1.19 pups per litter, suggesting overdominance as a possible mode of gene action in some cases. The observation of pleiotropic or linked QTL for growth and prolificacy corresponds well with results from many selection experiments identifying positively correlated responses to selection for these two traits. Received: 9 August 1997 / Accepted: 30 September 1997     

Advances in quantitative trait analysis in yeast

Understanding the genetic mechanisms underlying complex traits is one of the next frontiers in biology. The budding yeast Saccharomyces cerevisiae has become an important model for elucidating the mechanisms that govern natural genetic and phenotypic variation. This success is partially due to its intrinsic biological features, such as the short sexual generation time, high meiotic recombination rate, and small genome size. Precise reverse genetics technologies allow the high throughput manipulation of genetic information with exquisite precision, offering the unique opportunity to experimentally measure the phenotypic effect of genetic variants. Population genomic and phenomic studies have revealed widespread variation between diverged populations, characteristic of man-made environments, as well as geographic clusters of wild strains along with naturally occurring recombinant strains (mosaics). Here, we review these recent studies and provide a perspective on how these previously unappreciated levels of variation can help to bridge our understanding of the genotype-phenotype gap, keeping budding yeast at the forefront of genetic studies. Not only are quantitative trait loci (QTL) being mapped with high resolution down to the nucleotide, for the first time QTLs of modest effect and complex interactions between these QTLs and between QTLs and the environment are being determined experimentally at unprecedented levels using next generation techniques of deep sequencing selected pools of individuals as well as multi-generational crosses.     

A quantitative trait locus on chromosome 1 modulates intermale aggression in mice

Aggression between male conspecifics is a complex social behavior that is likely modulated by multiple gene variants. In this study, the BXD recombinant inbred mouse strains (RIS) were used to map quantitative trait loci (QTLs) underlying behaviors associated with intermale aggression. Four hundred and fifty‐seven males from 55 strains (including the parentals) were observed at an age of 13 ± 1 week in a resident‐intruder test following 10 days of isolation. Attack latency was measured directly within a 10‐minute time period and the test was repeated 24 hours later. The variables we analyzed were the proportion of attacking males in a given strain as well as the attack latency (on days 1 and 2, and both days combined). On day 1, 29% of males attacked, and this increased to 37% on day 2. Large strain differences were obtained for all measures of aggression, indicating substantial heritability (intraclass correlations 0.10‐0.18). We identified a significant QTL on chromosome (Chr) 1 and suggestive QTLs on mouse Chrs 1 and 12 for both attack and latency variables. The significant Chr 1 locus maps to a gene‐sparse region between 82 and 88.5 Mb with the C57BL/6J allele increasing aggression and explaining about 18% of the variance. The most likely candidate gene modulating this trait is Htr2b which encodes the serotonin 2B receptor and has been implicated in aggressive and impulsive behavior in mice, humans and other species.     

Multivariate analysis of quantitative trait loci influencing variation in anxiety-related behavior in laboratory mice

The number and mode of action of quantitative trait loci (QTL) that contribute to behavioral variation in rodents is still largely unknown. On theoretical grounds, multivariate techniques are expected to yield new insights into this problem, but there are only a few examples of its application in practice. Here we explore the power of multivariate approaches to uncover the genetic architecture of 23 anxiety-related phenotypes in 1636 F₂ laboratory mice. We detected QTL with a genome-wide significance threshold of P < 0.05 on 14 chromosomes, of which 10 correspond to those identified by univariate analysis. Novel QTL were found on Chromosomes 3, 9, 13, and 17. Thus, multivariate analyses increased the yield of QTL exceeding a genome-wide significance threshold by 40%. On the basis of these results and by the application of a QTL estimator, we show that the mean number of QTL influencing anxiety-related behavior in mice is 6, with a 95% upper limit of 14.     

A novel screening assay for hydroxynitrile lyases suitable for high-throughput screening

Hydroxynitrile lyases (Hnls) are important biocatalysts for the synthesis of optically pure cyanohydrins, which are used as precursors and building blocks for a wide range of high price fine chemicals. Although two Hnl enzymes, from the tropical rubber tree Hevea brasiliensis and from the almond tree Prunus amygdalus, are already used for large scale industrial applications, the enzymes still need to be improved and adapted to the special demands of industrial processes. In many cases directed evolution has been the method of choice to improve enzymes, which are applied as industrial biocatalysts. The screening procedure is the most crucial point in every directed evolution experiment. Herein, we describe the successful development of a novel screening assay for Hnls and its application in high-throughput screening of Escherichia coli mutant libraries. The new assay allows rapid screening of mutant libraries and facilitates the discovery of improved enzyme variants. Hnls catalyze the cleavage of cyanohydrins to hydrocyanic acid and the corresponding aldehyde or ketone. The enzyme assay is based on the detection of hydrocyanic acid produced, making it an all-purpose screening assay, without restriction to any kind of substrate. The gaseous HCN liberated within the Hnl reaction is detected by a visible colorimetric reaction. The facile, highly sensitive and reproducible screening method was validated by identifying new enzyme variants with novel substrate specificities.     

A statistical framework for quantitative trait mapping

We describe a general statistical framework for the genetic analysis of quantitative trait data in inbred line crosses. Our main result is based on the observation that, by conditioning on the unobserved QTL genotypes, the problem can be split into two statistically independent and manageable parts. The first part involves only the relationship between the QTL and the phenotype. The second part involves only the location of the QTL in the genome. We developed a simple Monte Carlo algorithm to implement Bayesian QTL analysis. This algorithm simulates multiple versions of complete genotype information on a genomewide grid of locations using information in the marker genotype data. Weights are assigned to the simulated genotypes to capture information in the phenotype data. The weighted complete genotypes are used to approximate quantities needed for statistical inference of QTL locations and effect sizes. One advantage of this approach is that only the weights are recomputed as the analyst considers different candidate models. This device allows the analyst to focus on modeling and model comparisons. The proposed framework can accommodate multiple interacting QTL, nonnormal and multivariate phenotypes, covariates, missing genotype data, and genotyping errors in any type of inbred line cross. A software tool implementing this procedure is available. We demonstrate our approach to QTL analysis using data from a mouse backcross population that is segregating multiple interacting QTL associated with salt-induced hypertension.