Factorial study of the RNA-seq computational workflow identifies biases as technical gene signatures

RNA-seq is a modular experimental and computational approach aiming in identifying and quantifying RNA molecules. The modularity of the RNA-seq technology enables adaptation of the protocol to develop new ways to explore RNA biology, but this modularity also brings forth the importance of methodological thoroughness. Liberty of approach comes with the responsibility of choices, and such choices must be informed. Here, we present an approach that identifies gene group-specific quantification biases in current RNA-seq software and references by processing datasets using diverse RNA-seq computational pipelines, and by decomposing these expression datasets with an independent component analysis matrix factorization method. By exploring the RNA-seq pipeline using this systemic approach, we identify genome annotations as a design choice that affects to the same extent quantification results as does the choice of aligners and quantifiers. We also show that the different choices in RNA-seq methodology are not independent, identifying interactions between genome annotations and quantification software. Genes were mainly affected by differences in their sequence, by overlapping genes and genes with similar sequence. Our approach offers an explanation for the observed biases by identifying the common features used differently by the software and references, therefore providing leads for the betterment of RNA-seq methodology.

Current RNA-seq methodology reporting limits reproducibility

Ribonucleic acid sequencing (RNA-seq) identifies and quantifies RNA molecules from a biological sample. Transformation from raw sequencing data to meaningful gene or isoform counts requires an in silico bioinformatics pipeline. Such pipelines are modular in nature, built using selected software and biological references. Software is usually chosen and parameterized according to the sequencing protocol and biological question. However, while biological and technical noise is alleviated through replicates, biases due to the pipeline and choice of biological references are often overlooked. Here, we show that the current standard practice prevents reproducibility in RNA-seq studies by failing to specify required methodological information. Peer-reviewed articles are intended to apply currently accepted scientific and methodological standards. Inasmuch as the bias-less and optimal RNA-seq pipeline is not perfectly defined, methodological information holds a meaningful role in defining the results. This work illustrates the need for a standardized and explicit display of methodological information in RNA-seq experiments.

Handling multi-mapped reads in RNA-seq

Many eukaryotic genomes harbour large numbers of duplicated sequences, of diverse biotypes, resulting from several mechanisms including recombination, whole genome duplication and retro-transposition. Such repeated sequences complicate gene/transcript quantification during RNA-seq analysis due to reads mapping to more than one locus, sometimes involving genes embedded in other genes. Genes of different biotypes have dissimilar levels of sequence duplication, with long-noncoding RNAs and messenger RNAs sharing less sequence similarity to other genes than biotypes encoding shorter RNAs. Many strategies have been elaborated to handle these multi-mapped reads, resulting in increased accuracy in gene/transcript quantification, although separate tools are typically used to estimate the abundance of short and long genes due to their dissimilar characteristics. This review discusses the mechanisms leading to sequence duplication, the biotypes affected, the computational strategies employed to deal with multi-mapped reads and the challenges that still remain to be overcome.

Human Hepatocyte Nuclear Factor 4-α Encodes Isoforms with Distinct Transcriptional Functions

HNF4α is a nuclear receptor produced as 12 isoforms from two promoters by alternative splicing. To characterize the transcriptional capacities of all 12 HNF4α isoforms, stable lines expressing each isoform were generated. The entire transcriptome associated with each isoform was analyzed as well as their respective interacting proteome. Major differences were noted in the transcriptional function of these isoforms. The α1 and α2 isoforms were the strongest regulators of gene expression whereas the α3 isoform exhibited significantly reduced activity. The α4, α5, and α6 isoforms, which use an alternative first exon, were characterized for the first time, and showed a greatly reduced transcriptional potential with an inability to recognize the consensus response element of HNF4α. Several transcription factors and coregulators were identified as potential specific partners for certain HNF4α isoforms. An analysis integrating the vast amount of omics data enabled the identification of transcriptional regulatory mechanisms specific to certain HNF4α isoforms, hence demonstrating the importance of considering all isoforms given their seemingly diverse functions.

Application of a new method for simultaneous phase and baseline correction of NMR signals (SINC)

Recently, we presented a new approach for simultaneous phase and baseline correction of nuclear magnetic resonance (NMR) signals (SINC) that is based on multiobjective optimization. The algorithm can automatically correct large sets of NMR spectra, which are commonly acquired when reactions and processes are monitored with NMR spectroscopy. The aim of the algorithm is to provide spectra that can be evaluated quantitatively, for example, to calculate the composition of a mixture or the extent of reaction. In this work, the SINC algorithm is tested in three different studies. In an in-house comparison study, spectra of different mixtures were corrected both with the SINC method and manually by different experienced users. The study shows that the results of the different users vary significantly and that their average uncertainty of the composition measurement is larger than the uncertainty obtained when the spectra are corrected with the SINC method. By means of a dilution study, we demonstrate that the SINC method is also applicable for the correction of spectra with low signal-to-noise ratio. Furthermore, a large set of NMR spectra that was acquired to follow a reaction was corrected with the SINC method. Even in this system, where the areas of the peaks and their chemical shifts changed during the course of reaction, the SINC method corrected the spectra robustly. The results show that this method is especially suited to correct large sets of NMR spectra and it is thus an important contribution for the automation of the evaluation of NMR spectra.

Current RNA-seq methodology reporting limits reproducibility

Ribonucleic acid sequencing (RNA-seq) identifies and quantifies RNA molecules from a biological sample. Transformation from raw sequencing data to meaningful gene or isoform counts requires an in silico bioinformatics pipeline. Such pipelines are modular in nature, built using selected software and biological references. Software is usually chosen and parameterized according to the sequencing protocol and biological question. However, while biological and technical noise is alleviated through replicates, biases due to the pipeline and choice of biological references are often overlooked. Here, we show that the current standard practice prevents reproducibility in RNA-seq studies by failing to specify required methodological information. Peer-reviewed articles are intended to apply currently accepted scientific and methodological standards. Inasmuch as the bias-less and optimal RNA-seq pipeline is not perfectly defined, methodological information holds a meaningful role in defining the results. This work illustrates the need for a standardized and explicit display of methodological information in RNA-seq experiments.

Multi-objective optimization for an automated and simultaneous phase and baseline correction of NMR spectral data

Spectral data preprocessing is an integral and sometimes inevitable part of chemometric analyses. For Nuclear Magnetic Resonance (NMR) spectra a possible first preprocessing step is a phase correction which is applied to the Fourier transformed free induction decay (FID) signal. This preprocessing step can be followed by a separate baseline correction step. Especially if series of high-resolution spectra are considered, then automated and computationally fast preprocessing routines are desirable. A new method is suggested that applies the phase and the baseline corrections simultaneously in an automated form without manual input, which distinguishes this work from other approaches. The underlying multi-objective optimization or Pareto optimization provides improved results compared to consecutively applied correction steps. The optimization process uses an objective function which applies strong penalty constraints and weaker regularization conditions. The new method includes an approach for the detection of zero baseline regions. The baseline correction uses a modified Whittaker smoother. The functionality of the new method is demonstrated for experimental NMR spectra. The results are verified against gravimetric data. The method is compared to alternative preprocessing tools. Additionally, the simultaneous correction method is compared to a consecutive application of the two correction steps.

Interaction of 5-HT2A and 5-HT2C receptors in R(-)-2,5-dimethoxy-4-iodoamphetamine-elicited head twitch behavior in mice

Drug-elicited head-twitch behavior is a useful model for studying hallucinogen activity at 5-HT(2A) receptors in the mouse. Chemically diverse compounds active in this assay yield biphasic dose-effect curves, but there is no compelling explanation for the "descending" portion of these functions. A set of experiments was designed to test the hypothesis that the induction of head-twitch behavior is mediated by agonist actions at 5-HT(2A) receptors, whereas the inhibition of head-twitch behavior observed at higher doses results from competing agonist activity at 5-HT(2C) receptors. The effects of the phenethylamine hallucinogen R(-)-2,5-dimethoxy-4-iodoamphetamine (DOI) on head-twitch behavior were studied over a range of doses in the mouse, generating a characteristic biphasic dose-response curve. Pretreatment with the selective 5-HT(2A) antagonist (+)-(2,3-dimethoxyphenyl)-1-[2-(4-fluorophenylethyl)]-4-piperidine-methanol (M100907) shifted only the ascending limb of the DOI dose-effect function, whereas pretreatment with the nonselective 5-HT(2A/2C) antagonist 3-{2-[4-(4-fluorobenzoyl)piperidin-1-yl]ethyl}quinazoline-2,4(1H,3H)-dione (ketanserin) produced a parallel shift to the right in the DOI dose-response curve. Administration of the 5-HT(2C) agonist S-2-(chloro-5-fluoro-indol-l-yl)-1-methylethylamine (Ro 60-0175) noncompetitively inhibited DOI-elicited head-twitch behavior across the entire dose-effect function. Finally, pretreatment with the selective 5-HT(2C) antagonists 6-chloro-5-methyl-1-[(2-[2-methylpyrid-3-yloxy]pyrid-5yl)carbamoyl]indoline (SB242084) or 8-[5-(2,4-dimethoxy-5-(4-trifluoromethylphenylsulfonamido)phenyl-5-oxopentyl]-1,3,8-triazaspiro[4,5]decane-2,4-dione hydrochloride (RS 102221) did not alter DOI-elicited head-twitch behavior on the ascending limb of the dose-response curve but shifted the descending limb of the DOI dose-response function to the right. The results of these experiments provide strong evidence that DOI-elicited head-twitch behavior is a 5-HT(2A) agonist-mediated effect, with subsequent inhibition of head-twitch behavior being driven by competing 5-HT(2C) agonist activity.

Distress, denial, and low adherence to behavioral interventions predict faster disease progression in gay men infected with human immunodeficiency virus

This study examined psychological prediction of 2-year disease progression in gay men after finding out their human immunodeficiency virus (HIV) serostatus. Psychological and immune status of asymptomatic gay men who did not know their HIV serostatus was monitored during the 5 weeks before and after serostatus notification. The men were randomly assigned to an exercise. cognitive-behavioral stress-management intervention, or control group. At 2-year follow-up for the 23 men who turned out to be seropositive. 9 had developed symptoms, including 5 with acquired immune deficiency syndrome--4 of whom died. Distress at diagnosis, denial (5 weeks post-diagnosis minus pre-diagnosis). and low adherence during interventions were significant predictors of 2-year disease progression. Denial and adherence remained significant predictors of disease progression even after controlling for CD4 number at entry. Furthermore. change in denial was significantly correlated with immune status 1 year later; l-year immune status was significantly correlated with 2-year disease progression. The present study therefore demonstrates significant relations between psychological variables on the one hand and both immune measures and HIV-1 disease progression on the other. We conclude that distress, denial, and low protocol compliance predict subsequent disease progression.

Sleep disturbances in men with asymptomatic human immunodeficiency (HIV) infection

During the clinical latency phase of human immunodeficiency virus (HIV) disease the central nervous system may be infected and begin to manifest subtle dysfunction. Our early investigations demonstrated persistent alterations in the sleep architecture of HIV-infected asymptomatic men. The major aims of this study were to delineate alterations of sleep architecture in asymptomatic HIV-infected men, to identify and describe sleep behavior complaints and to seek a correlation between objective sleep parameters and subjective complaints of sleep behavior. The study sample consisted of 24 men, 14 HIV-infected and 10 HIV-negative, age-matched controls. The protocol included a comprehensive history and physical, two polysomnograms, urine toxicity, detailed written sleep questionnaire, the Pittsburgh Sleep Quality Index, the Spielberger State-Trait Anxiety Test and the Beck Depression Inventory. Our results indicated that sleep architecture differed from controls in that wakefulness, slow-wave sleep [SWS-stage 3 and 4 nonrapid eye movement (NREM) sleep] and stage rapid eye movement (REM) sleep were more evenly dispersed throughout the night. In particular, SWS was prevalent during the second half of recorded sleep. The observed changes in the NREM/REM cycle could not be explained on the basis of underlying psychopathology. Just as the course of individuals with HIV infection varies, it is expected that sleep abnormalities will vary. Considering the known relationships between NREM stage 3 and 4 and immune system function, it is possible that the observed alterations in the NREM/REM cycle are related to coincident changes in immunologic function. Quantitative measures of NREM sleep, especially SWS and REM sleep, are perhaps of greater significance than relative measures of sleep stages.(ABSTRACT TRUNCATED AT 250 WORDS)

Immunologic function in a cohort of human immunodeficiency virus type 1-seropositive and -negative healthy homosexual men

The study objectives were to determine the early effects of human immunodeficiency virus type 1 (HIV-1) infection on both phenotypic and functional immunologic markers in healthy homosexual men, to ascertain the relationships of these markers to each other, and to discover which markers were affected by enrollment in an AIDS study in which HIV-1 serostatus would be determined. The major findings were as follows. (i) The CD4/CD8 ratio and lymphocyte proliferative response to pokeweed mitogen were the characteristics most affected by early HIV-1 infection. (ii) The loss in CD4 cells observed in the HIV-1-positive homosexual men was entirely due to diminished numbers of the memory subset. CD4+ CD29+. The reciprocal subset of CD4, CD4+ CD45RA+, did not differ in the two groups of homosexual men at either time point or in the controls. (iii) Prior to learning their HIV-1 serostatus, HIV-1 antibody-negative risk-group males had lower phytohemagglutinin (PHA) responses than the controls did. In the assays following notification of their seronegativity, however, these men had PHA values which were not different from those of the controls. In the HIV-1-positive group, the responses to both PHA and pokeweed mitogen were below those of both HIV-1-negative groups and did not change after serostatus notification. (iv) The activity of natural killer cells was lower in the risk-group men than in the controls at both pre- and postdiagnosis but was not related to HIV-1 serostatus. (v) In this cohort of homosexual men, the CD4/CD8 ratio correlated significantly with the functional measures of immunologic status in the HIV-1-positive men, but not in the HIV-1-negative men.