Effects of filtering by Present call on analysis of microarray experiments

Background

Affymetrix GeneChips^® are widely used for expression profiling of tens of thousands of genes. The large number of comparisons can lead to false positives. Various methods have been used to reduce false positives, but they have rarely been compared or quantitatively evaluated. Here we describe and evaluate a simple method that uses the detection (Present/Absent) call generated by the Affymetrix microarray suite version 5 software (MAS5) to remove data that is not reliably detected before further analysis, and compare this with filtering by expression level. We explore the effects of various thresholds for removing data in experiments of different size (from 3 to 10 arrays per treatment), as well as their relative power to detect significant differences in expression.

Results

Our approach sets a threshold for the fraction of arrays called Present in at least one treatment group. This method removes a large percentage of probe sets called Absent before carrying out the comparisons, while retaining most of the probe sets called Present. It preferentially retains the more significant probe sets (p ≤ 0.001) and those probe sets that are turned on or off, and improves the false discovery rate. Permutations to estimate false positives indicate that probe sets removed by the filter contribute a disproportionate number of false positives. Filtering by fraction Present is effective when applied to data generated either by the MAS5 algorithm or by other probe-level algorithms, for example RMA (robust multichip average). Experiment size greatly affects the ability to reproducibly detect significant differences, and also impacts the effect of filtering; smaller experiments (3–5 samples per treatment group) benefit from more restrictive filtering (≥50% Present).

Conclusion

Use of a threshold fraction of Present detection calls (derived by MAS5) provided a simple method that effectively eliminated from analysis probe sets that are unlikely to be reliable while preserving the most significant probe sets and those turned on or off; it thereby increased the ratio of true positives to false positives.     

Extensive up-regulation of gene expression in cancer: the normalised use of microarray data

Based on the assumption that only a few genes are differentially expressed in a disease and have balanced upward and downward expression level changes, researchers usually normalise microarray data by forcing all of the arrays to have the same probe intensity distributions to remove technical variations in the data. However, accumulated evidence suggests that gene expressions could be widely altered in cancer, so we need to evaluate the sensitivities of biological discoveries to violation of the normalisation assumption. Here, we show that the medians of the original probe intensities increase in most of the ten cancer types analyzed in this paper, indicating that genes may be widely up-regulated in many cancer types. Thus, at least for cancer study, normalising all arrays to have the same distribution of probe intensities regardless of the state (diseased vs. normal) tends to falsely produce many down-regulated differentially expressed (DE) genes while missing many truly up-regulated DE genes. We also show that the DE genes solely detected in the non-normalised data for cancers are highly reproducible across different datasets for the same cancers, indicating that effective biological signals naturally exist in the non-normalised data. Because the powers of current statistical analyses using the non-normalised data tend to be low, we suggest selecting DE genes in both normalised and non-normalised data and then filter out the false DE genes extracted from the normalised data that show opposite deregulation directions in the non-normalised data.     

Bayesian Joint Analysis of Gene Expression Data and Gene Functional Annotations

Identifying which genes and which gene sets are differentially expressed (DE) under two experimental conditions are both key questions in microarray analysis. Although closely related and seemingly similar, they cannot replace each other, due to their own importance and merits in scientific discoveries. Existing approaches have been developed to address only one of the two questions. Further, most of the methods for detecting DE genes purely rely on gene expression analysis, without using the information about gene functional grouping. Methods for detecting altered gene sets often use a two-step procedure, of which the first step conducts differential expression analysis using expression data only, and the second step takes results from the first step and tries to examine whether each predefined gene set is overrepresented by DE genes through some testing procedure. Such a sequential manner in analysis might cause information loss by just focusing on summary results without using the entire expression data in the second step. Here, we propose a Bayesian joint modeling approach to address the two key questions in parallel, which incorporates the information of functional annotations into expression data analysis and meanwhile infer the enrichment of functional groups. Simulation results and analysis of experimental data obtained for E.?coli show improved statistical power of our integrated approach in both identifying DE genes and altered gene sets, when compared to conventional methods.     

Identifying Differentially Expressed Genes in Time Course Microarray Data

Identifying differentially expressed (DE) genes across conditions or treatments is a typical problem in microarray experiments. In time course microarray experiments (under two or more conditions/treatments), it is sometimes of interest to identify two classes of DE genes: those with no time-condition interactions (called parallel DE genes, or PDE), and those with time-condition interactions (nonparallel DE genes, NPDE). Although many methods have been proposed for identifying DE genes in time course experiments, methods for discerning NPDE genes from the general DE genes are still lacking. We propose a functional ANOVA mixed-effect model to model time course gene expression observations. The fixed effect of (the mean curve) of the model decomposes bivariate functions of time and treatments (or experimental conditions) as in the classic ANOVA method and provides the associated notions of main effects and interactions. Random effects capture time-dependent correlation structures. In this model, identifying NPDE genes is equivalent to testing the significance of the time-condition interaction, for which an approximate F-test is suggested. We examined the performance of the proposed method on simulated datasets in comparison with some existing methods, and applied the method to a study of human reaction to the endotoxin stimulation, as well as to a cell cycle expression data set.     

I/NI-calls for the exclusion of non-informative genes: a highly effective filtering tool for microarray data

MOTIVATION: DNA microarray technology typically generates many measurements of which only a relatively small subset is informative for the interpretation of the experiment. To avoid false positive results, it is therefore critical to select the informative genes from the large noisy data before the actual analysis. Most currently available filtering techniques are supervised and therefore suffer from a potential risk of overfitting. The unsupervised filtering techniques, on the other hand, are either not very efficient or too stringent as they may mix up signal with noise. We propose to use the multiple probes measuring the same target mRNA as repeated measures to quantify the signal-to-noise ratio of that specific probe set. A Bayesian factor analysis with specifically chosen prior settings, which models this probe level information, is providing an objective feature filtering technique, named informative/non-informative calls (I/NI calls). RESULTS: Based on 30 real-life data sets (including various human, rat, mice and Arabidopsis studies) and a spiked-in data set, it is shown that I/NI calls is highly effective, with exclusion rates ranging from 70% to 99%. Consequently, it offers a critical solution to the curse of high-dimensionality in the analysis of microarray data. AVAILABILITY: This filtering approach is publicly available as a function implemented in the R package FARMS (www.bioinf.jku.at/software/farms/farms.html).     

Detection of differentially expressed gene sets in a partially paired microarray data set

Partially paired data sets often occur in microarray experiments (Kim et al., 2005; Liu, Liang and Jang, 2006). Discussions of testing with partially paired data are found in the literature (Lin and Stivers 1974; Ekbohm, 1976; Bhoj, 1978). Bhoj (1978) initially proposed a test statistic that uses a convex combination of paired and unpaired t statistics. Kim et al. (2005) later proposed the t3 statistic, which is a linear combination of paired and unpaired t statistics, and then used it to detect differentially expressed (DE) genes in colorectal cancer (CRC) cDNA microarray data. In this paper, we extend Kim et al.'s t3 statistic to the Hotelling's T2 type statistic Tp for detecting DE gene sets of size p. We employ Efron's empirical null principle to incorporate inter-gene correlation in the estimation of the false discovery rate. Then, the proposed Tp statistic is applied to Kim et al's CRC data to detect the DE gene sets of sizes p=2 and p=3. Our results show that for small p, particularly for p=2 and marginally for p=3, the proposed Tp statistic compliments the univariate procedure by detecting additional DE genes that were undetected in the univariate test procedure. We also conduct a simulation study to demonstrate that Efron's empirical null principle is robust to the departure from the normal assumption.     

miBLAST: scalable evaluation of a batch of nucleotide sequence queries with BLAST

A common task in many modern bioinformatics applications is to match a set of nucleotide query sequences against a large sequence dataset. Exis-ting tools, such as BLAST, are designed to evaluate a single query at a time and can be unacceptably slow when the number of sequences in the query set is large. In this paper, we present a new algorithm, called miBLAST, that evaluates such batch workloads efficiently. At the core, miBLAST employs a q-gram filtering and an index join for efficiently detecting similarity between the query sequences and database sequences. This set-oriented technique, which indexes both the query and the database sets, results in substantial performance improvements over existing methods. Our results show that miBLAST is significantly faster than BLAST in many cases. For example, miBLAST aligned 247965 oligonucleotide sequences in the Affymetrix probe set against the Human UniGene in 1.26 days, compared with 27.27 days with BLAST (an improvement by a factor of 22). The relative performance of miBLAST increases for larger word sizes; however, it decreases for longer queries. miBLAST employs the familiar BLAST statistical model and output format, guaranteeing the same accuracy as BLAST and facilitating a seamless transition for existing BLAST users.     

Fitting hidden Markov models of protein domains to a target species: application to Plasmodium falciparum

Background

The identification of gene sets that are significantly impacted in a given condition based on microarray data is a crucial step in current life science research. Most gene set analysis methods treat genes equally, regardless how specific they are to a given gene set.

Results

In this work we propose a new gene set analysis method that computes a gene set score as the mean of absolute values of weighted moderated gene t-scores. The gene weights are designed to emphasize the genes appearing in few gene sets, versus genes that appear in many gene sets. We demonstrate the usefulness of the method when analyzing gene sets that correspond to the KEGG pathways, and hence we called our method P athway A nalysis with D own-weighting of O verlapping G enes (PADOG). Unlike most gene set analysis methods which are validated through the analysis of 2-3 data sets followed by a human interpretation of the results, the validation employed here uses 24 different data sets and a completely objective assessment scheme that makes minimal assumptions and eliminates the need for possibly biased human assessments of the analysis results.

Conclusions

PADOG significantly improves gene set ranking and boosts sensitivity of analysis using information already available in the gene expression profiles and the collection of gene sets to be analyzed. The advantages of PADOG over other existing approaches are shown to be stable to changes in the database of gene sets to be analyzed. PADOG was implemented as an R package available at: http://bioinformaticsprb.med.wayne.edu/PADOG/or http://www.bioconductor.org.     

Determination of the differentially expressed genes in microarray experiments using local FDR

Background  

Thousands of genes in a genomewide data set are tested against some null hypothesis, for detecting differentially expressed genes in microarray experiments. The expected proportion of false positive genes in a set of genes, called the False Discovery Rate (FDR), has been proposed to measure the statistical significance of this set. Various procedures exist for controlling the FDR. However the threshold (generally 5%) is arbitrary and a specific measure associated with each gene would be worthwhile.     

Analysis of a simulated microarray dataset: Comparison of methods for data normalisation and detection of differential expression (Open Access publication)

Microarrays allow researchers to measure the expression of thousands of genes in a single experiment. Before statistical comparisons can be made, the data must be assessed for quality and normalisation procedures must be applied, of which many have been proposed. Methods of comparing the normalised data are also abundant, and no clear consensus has yet been reached. The purpose of this paper was to compare those methods used by the EADGENE network on a very noisy simulated data set. With the a priori knowledge of which genes are differentially expressed, it is possible to compare the success of each approach quantitatively. Use of an intensity-dependent normalisation procedure was common, as was correction for multiple testing. Most variety in performance resulted from differing approaches to data quality and the use of different statistical tests. Very few of the methods used any kind of background correction. A number of approaches achieved a success rate of 95% or above, with relatively small numbers of false positives and negatives. Applying stringent spot selection criteria and elimination of data did not improve the false positive rate and greatly increased the false negative rate. However, most approaches performed well, and it is encouraging that widely available techniques can achieve such good results on a very noisy data set.     

Essential molecular functions associated with the circular code evolution

A circular code is a set of trinucleotides allowing the reading frames in genes to be retrieved locally, i.e. anywhere in genes and in particular without start codons, and automatically with a window of few nucleotides. In 1996, a common circular code, called X, was identified in large populations of eukaryotic and prokaryotic genes. Hence, it is believed to be an ancestral structural property of genes. A new computational approach based on comparative genomics is developed to identify essential molecular functions associated with circular codes. It is based on a quantitative and sensitive statistical method (FPTF) to identify three permuted trinucleotide sets in the three frames of genes, a flower automaton algorithm to determine if a trinucleotide set is a circular code or not, and an integrated Gene Ontology and Taxonomy (iGOT) database. By carrying out automatic circular code analyses on a huge number of gene populations where each population is associated with a particular molecular function, it identifies 266 gene populations having circular codes close to X. Surprisingly, their molecular functions include 98% of those covered by the essential genes of the DEG database (Database of Essential Genes). Furthermore, three trinucleotides GTG, AAG and GCG, replacing three trinucleotides of the code X and called “evolutionary” trinucleotides, significantly occur in these 266 gene populations. Finally, a new method developed to analyse and quantify the stability of a set of trinucleotides demonstrates that these evolutionary trinucleotides are associated with a significant increase of the stability of the common circular code X. Indeed, its stability increases from the 1502th rank to the 16th rank after the replacement of the three evolutionary trinucleotides among 9920 possible trinucleotide replacement sets.     

Identifying Differentially Expressed Genes for Time-course Microarray Data through Functional Data Analysis

Identification of differentially expressed (DE) genes across two conditions is a common task with microarray. Most existing approaches accomplish this goal by examining each gene separately based on a model and then control the false discovery rate over all genes. We took a different approach that employs a uniform platform to simultaneously depict the dynamics of the gene trajectories for all genes and select differently expressed genes. A new Functional Principal Component (FPC) approach is developed for time-course microarray data to borrow strength across genes. The approach is flexible as the temporal trajectory of the gene expressions is modeled nonparametrically through a set of orthogonal basis functions, and often fewer basis functions are needed to capture the shape of the gene expression trajectory than existing nonparametric methods. These basis functions are estimated from the data reflecting major modes of variation in the data. The correlation structure of the gene expressions over time is also incorporated without any parametric assumptions and estimated from all genes such that the information across other genes can be shared to infer one individual gene. Estimation of the parameters is carried out by an efficient hybrid EM algorithm. The performance of the proposed method across different scenarios was compared favorably in simulation to two-way mixed-effects ANOVA and the EDGE method using B-spline basis function. Application to the real data on C. elegans developmental stages also suggested that FPC analysis combined with hybrid EM algorithm provides a computationally fast and efficient method for identifying DE genes based on time-course microarray data.     

Genome-wide co-expression based prediction of differential expressions

MOTIVATION: Microarrays have been widely used for medical studies to detect novel disease-related genes. They enable us to study differential gene expressions at a genomic level. They also provide us with informative genome-wide co-expressions. Although many statistical methods have been proposed for identifying differentially expressed genes, genome-wide co-expressions have not been well considered for this issue. Incorporating genome-wide co-expression information in the differential expression analysis may improve the detection of disease-related genes. RESULTS: In this study, we proposed a statistical method for predicting differential expressions through the local regression between differential expression and co-expression measures. The smoother span parameter was determined by optimizing the rank correlation between the observed and predicted differential expression measures. A mixture normal quantile-based method was used to transform data. We used the gene-specific permutation procedure to evaluate the significance of a prediction. Two published microarray data sets were analyzed for applications. For the data set collected for a prostate cancer study, the proposed method identified many genes with weak differential expressions. Several of these genes have been shown in literature to be associated with the disease. For the data set collected for a type 2 diabetes study, no significant genes could be identified by the traditional methods. However, the proposed method identified many genes with significantly low false discovery rates. AVAILABILITY: The R codes are freely available at http://home.gwu.edu/~ylai/research/CoDiff, where the gene lists ranked by our method are also provided as the Supplementary Material.     

A Condition-Enumeration Tree method for mining biclusters from DNA microarray data sets

Biclustering, which performs simultaneous clustering of rows (e.g., genes) and columns (e.g., conditions), has proved of great value for finding interesting patterns from microarray data. To find biclusters, a model called pCluster was proposed. A pCluster consists of a set of genes and a set of conditions, where the expression levels of these genes have a similar variation under these conditions. Based on this model, most of the previous methods need to compute MDSs (maximum dimension sets) for every two genes in the microarray data. Since the number of genes is far larger than the number of conditions, this step is inefficient. Another method called MicroCluster was proposed. This method does not compute MDSs for every two genes, and transforms the problem into a graph problem. However, it needs to solve the Maximal Clique problem, which is NP-Complete. To avoid the above disadvantages, in this paper, we propose a new method, CE-Tree (Condition-Enumeration Tree), for finding pClusters. Instead of generating MDSs for every two genes, we generate only MDSs for every two conditions. Then, based only on these MDSs, we expand the CE-Tree in a special local breadth-first within global depth-first manner to efficiently find all pClusters. We also utilize the idea of the traditional hash join approach to efficiently support the CE-Tree. From the simulation results, we show that the CE-Tree method could find pClusters more efficiently than those previous methods.     

An empirical Bayesian approach for identifying differential coexpression in high-throughput experiments

A common goal of microarray and related high-throughput genomic experiments is to identify genes that vary across biological condition. Most often this is accomplished by identifying genes with changes in mean expression level, so called differentially expressed (DE) genes, and a number of effective methods for identifying DE genes have been developed. Although useful, these approaches do not accommodate other types of differential regulation. An important example concerns differential coexpression (DC). Investigations of this class of genes are hampered by the large cardinality of the space to be interrogated as well as by influential outliers. As a result, existing DC approaches are often underpowered, exceedingly prone to false discoveries, and/or computationally intractable for even a moderately large number of pairs. To address this, an empirical Bayesian approach for identifying DC gene pairs is developed. The approach provides a false discovery rate controlled list of significant DC gene pairs without sacrificing power. It is applicable within a single study as well as across multiple studies. Computations are greatly facilitated by a modification to the expectation-maximization algorithm and a procedural heuristic. Simulations suggest that the proposed approach outperforms existing methods in far less computational time; and case study results suggest that the approach will likely prove to be a useful complement to current DE methods in high-throughput genomic studies.     

Large scale maximum average power multiple inference on time-course count data with application to RNA-seq analysis

Experiments that longitudinally collect RNA sequencing (RNA-seq) data can provide transformative insights in biology research by revealing the dynamic patterns of genes. Such experiments create a great demand for new analytic approaches to identify differentially expressed (DE) genes based on large-scale time-course count data. Existing methods, however, are suboptimal with respect to power and may lack theoretical justification. Furthermore, most existing tests are designed to distinguish among conditions based on overall differential patterns across time, though in practice, a variety of composite hypotheses are of more scientific interest. Finally, some current methods may fail to control the false discovery rate. In this paper, we propose a new model and testing procedure to address the above issues simultaneously. Specifically, conditional on a latent Gaussian mixture with evolving means, we model the data by negative binomial distributions. Motivated by Storey (2007) and Hwang and Liu (2010), we introduce a general testing framework based on the proposed model and show that the proposed test enjoys the optimality property of maximum average power. The test allows not only identification of traditional DE genes but also testing of a variety of composite hypotheses of biological interest. We establish the identifiability of the proposed model, implement the proposed method via efficient algorithms, and demonstrate its good performance via simulation studies. The procedure reveals interesting biological insights, when applied to data from an experiment that examines the effect of varying light environments on the fundamental physiology of the marine diatom Phaeodactylum tricornutum.