e-Cigarettes, Smoking Cessation, and Weight Change: Retrospective Secondary Analysis of the Evaluating the Efficacy of e-Cigarette Use for Smoking Cessation Trial.

BACKGROUND: While smoking cessation has been linked to substantial weight gain, the potential influence of e-cigarettes on weight changes among individuals who use these devices to quit smoking is not fully understood. OBJECTIVE: This study aims to reanalyze data from the Evaluating the Efficacy of e-Cigarette Use for Smoking Cessation (E3) trial to assess the causal effects of e-cigarette use on change in body weight. METHODS: This is a secondary analysis of the E3 trial in which participants were randomized into 3 groups: nicotine e-cigarettes plus counseling, nonnicotine e-cigarettes plus counseling, and counseling alone. With adjustment for baseline variables and the follow-up smoking abstinence status, weight changes were compared between the groups from baseline to 12 weeks' follow-up. Intention-to-treat and as-treated analyses were conducted using doubly robust estimation. Further causal analysis used 2 different propensity scoring methods to estimate causal regression curves for 4 smoking-related continuous variables. We evaluated 5 different subsets of data for each method. Selection bias was addressed, and missing data were imputed by the machine learning method extreme gradient boosting (XGBoost). RESULTS: A total of 257 individuals with measured weight at week 12 (mean age: 52, SD 12 y; women: n=122, 47.5%) were included. Across the 3 treatment groups, of the 257 participants, 204 (79.4%) who continued to smoke had, on average, largely unchanged weight at 12 weeks, with comparable mean weight gain ranging from -0.24 kg to 0.33 kg, while 53 (20.6%) smoking-abstinent participants gained weight, with a mean weight gain ranging from 2.05 kg to 2.70 kg. After adjustment, our analyses showed that the 2 e-cigarette arms exhibited a mean gain of 0.56 kg versus the counseling alone arm. The causal regression curves analysis also showed no strong evidence supporting a causal relationship between weight gain and the 3 e-cigarette-related variables. e-Cigarettes have small and variable causal effects on weight gain associated with smoking cessation. CONCLUSIONS: In the E3 trial, e-cigarettes seemed to have minimal effects on mitigating the weight gain observed in individuals who smoke and subsequently quit at 3 months. However, given the modest sample size and the potential underuse of e-cigarettes among those randomized to the e-cigarette treatment arms, these results need to be replicated in large, adequately powered trials. TRIAL REGISTRATION: ClinicalTrials.gov NCT02417467; https://www.clinicaltrials.gov/study/NCT02417467.

Toward an Explainable Large Language Model for the Automatic Identification of the Drug-Induced Liver Injury Literature.

Drug-induced liver injury (DILI) stands as a significant concern in drug safety, representing the primary cause of acute liver failure. Identifying the scientific literature related to DILI is crucial for monitoring, investigating, and conducting meta-analyses of drug safety issues. Given the intricate and often obscure nature of drug interactions, simple keyword searching can be insufficient for the exhaustive retrieval of the DILI-relevant literature. Manual curation of DILI-related publications demands pharmaceutical expertise and is susceptible to errors, severely limiting throughput. Despite numerous efforts utilizing cutting-edge natural language processing and deep learning techniques to automatically identify the DILI-related literature, their performance remains suboptimal for real-world applications in clinical research and regulatory contexts. In the past year, large language models (LLMs) such as ChatGPT and its open-source counterpart LLaMA have achieved groundbreaking progress in natural language understanding and question answering, paving the way for the automated, high-throughput identification of the DILI-related literature and subsequent analysis. Leveraging a large-scale public dataset comprising 14 203 training publications from the CAMDA 2022 literature AI challenge, we have developed what we believe to be the first LLM specialized in DILI analysis based on LLaMA-2. In comparison with other smaller language models such as BERT, GPT, and their variants, LLaMA-2 exhibits an enhanced out-of-fold accuracy of 97.19% and area under the ROC curve of 0.9947 using 3-fold cross-validation on the training set. Despite LLMs' initial design for dialogue systems, our study illustrates their successful adaptation into accurate classifiers for automated identification of the DILI-related literature from vast collections of documents. This work is a step toward unleashing the potential of LLMs in the context of regulatory science and facilitating the regulatory review process.

Automatic text classification of drug-induced liver injury using document-term matrix and XGBoost.

Introduction: Regulatory agencies generate a vast amount of textual data in the review process. For example, drug labeling serves as a valuable resource for regulatory agencies, such as U.S. Food and Drug Administration (FDA) and Europe Medical Agency (EMA), to communicate drug safety and effectiveness information to healthcare professionals and patients. Drug labeling also serves as a resource for pharmacovigilance and drug safety research. Automated text classification would significantly improve the analysis of drug labeling documents and conserve reviewer resources. Methods: We utilized artificial intelligence in this study to classify drug-induced liver injury (DILI)-related content from drug labeling documents based on FDA's DILIrank dataset. We employed text mining and XGBoost models and utilized the Preferred Terms of Medical queries for adverse event standards to simplify the elimination of common words and phrases while retaining medical standard terms for FDA and EMA drug label datasets. Then, we constructed a document term matrix using weights computed by Term Frequency-Inverse Document Frequency (TF-IDF) for each included word/term/token. Results: The automatic text classification model exhibited robust performance in predicting DILI, achieving cross-validation AUC scores exceeding 0.90 for both drug labels from FDA and EMA and literature abstracts from the Critical Assessment of Massive Data Analysis (CAMDA). Discussion: Moreover, the text mining and XGBoost functions demonstrated in this study can be applied to other text processing and classification tasks.

Breeding-assisted genomics: Applying meta-GWAS for milling and baking quality in CIMMYT wheat breeding program.

One of the biggest challenges for genetic studies on natural or unstructured populations is the unbalanced datasets where individuals are measured at different times and environments. This problem is also common in crop and animal breeding where many individuals are only evaluated for a single year and large but unbalanced datasets can be generated over multiple years. Many wheat breeding programs have focused on increasing bread wheat (Triticum aestivum L.) yield, but processing and end-use quality are critical components when considering its use in feeding the rising population of the next century. The challenges with end-use quality trait improvements are high cost and seed amounts for testing, the latter making selection in early breeding populations impossible. Here we describe a novel approach to identify marker-trait associations within a breeding program using a meta-genome wide association study (GWAS), which combines GWAS analysis from multi-year unbalanced breeding nurseries, in a manner reflecting meta-GWAS in humans. This method facilitated mapping of processing and end-use quality phenotypes from advanced breeding lines (n = 4,095) of the CIMMYT bread wheat breeding program from 2009 to 2014. Using the meta-GWAS we identified marker-trait associations, allele effects, candidate genes, and can select using markers generated in this process. Finally, the scope of this approach can be broadly applied in 'breeding-assisted genomics' across many crops to greatly increase our functional understanding of plant genomes.

Two approaches for estimating the lower limit of quantitation (LLOQ) of microRNA levels assayed as exploratory biomarkers by RT-qPCR.

BACKGROUND: Circulating microRNAs are undergoing exploratory use as safety biomarkers in drug development. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) is one common approach used to quantitate levels of microRNAs in samples that includes the use of a standard curve of calibrators fit to a regression model. Guidelines are needed for setting assay quantitation thresholds that are appropriate for this method and to biomarker pre-validation. RESULTS: In this report, we develop two workflows for determining a lower limit of quantitation (LLOQ) for RT-qPCR assays of microRNAs in exploratory studies. One workflow is based on an error threshold calculated by a logistic model of the calibration curve data. The second workflow is based on a threshold set by the sample blank, which is the no template control for RT-qPCR. The two workflows are used to set lower thresholds of reportable microRNA levels for an example dataset in which miR-208a levels in biofluids are quantitated in a cardiac injury model. LLOQ thresholds set by either workflow are effective in filtering out microRNA values with large uncertainty estimates. CONCLUSIONS: Two workflows for LLOQ determinations are presented in this report that provide methods that are easy to implement in investigational studies of microRNA safety biomarkers and offer choices in levels of conservatism in setting lower limits of acceptable values that facilitate interpretation of results.

Identification of dietary patterns associated with obesity in a nationally representative survey of Canadian adults: application of a priori, hybrid, and simplified dietary pattern techniques.

Background: Analyzing the effects of dietary patterns is an important approach for examining the complex role of nutrition in the etiology of obesity and chronic diseases.Objectives: The objectives of this study were to characterize the dietary patterns of Canadians with the use of a priori, hybrid, and simplified dietary pattern techniques, and to compare the associations of these patterns with obesity risk in individuals with and without chronic diseases (unhealthy and healthy obesity).Design: Dietary recalls from 11,748 participants (≥18 y of age) in the cross-sectional, nationally representative Canadian Community Health Survey 2.2 were used. A priori dietary pattern was characterized with the use of the previously validated 2015 Dietary Guidelines for Americans Adherence Index (DGAI). Weighted partial least squares (hybrid method) was used to derive an energy-dense (ED), high-fat (HF), low-fiber density (LFD) dietary pattern with the use of 38 food groups. The associations of derived dietary patterns with disease outcomes were then tested with the use of multinomial logistic regression.Results: An ED, HF, and LFD dietary pattern had high positive loadings for fast foods, carbonated drinks, and refined grains, and high negative loadings for whole fruits and vegetables (≥|0.17|). Food groups with a high loading were summed to form a simplified dietary pattern score. Moving from the first (healthiest) to the fourth (least healthy) quartiles of the ED, HF, and LFD pattern and the simplified dietary pattern scores was associated with increasingly elevated ORs for unhealthy obesity, with individuals in quartile 4 having an OR of 2.57 (95% CI: 1.75, 3.76) and 2.73 (95% CI: 1.88, 3.98), respectively (P-trend < 0.0001). Individuals who adhered the most to the 2015 DGAI recommendations (quartile 4) had a 53% lower OR of unhealthy obesity (P-trend < 0.0001). The associations of dietary patterns with healthy obesity and unhealthy nonobesity were weaker, albeit significant.Conclusions: Consuming an ED, HF, and LFD dietary pattern and lack of adherence to the recommendations of the 2015 DGAI were associated with a significantly higher risk of obesity with and without accompanying chronic diseases.

A DREAM Challenge to Build Prediction Models for Short-Term Discontinuation of Docetaxel in Metastatic Castration-Resistant Prostate Cancer.

PURPOSE: Docetaxel has a demonstrated survival benefit for patients with metastatic castration-resistant prostate cancer (mCRPC); however, 10% to 20% of patients discontinue docetaxel prematurely because of toxicity-induced adverse events, and the management of risk factors for toxicity remains a challenge. PATIENTS AND METHODS: The comparator arms of four phase III clinical trials in first-line mCRPC were collected, annotated, and compiled, with a total of 2,070 patients. Early discontinuation was defined as treatment stoppage within 3 months as a result of adverse treatment effects; 10% of patients discontinued treatment. We designed an open-data, crowd-sourced DREAM Challenge for developing models with which to predict early discontinuation of docetaxel treatment. Clinical features for all four trials and outcomes for three of the four trials were made publicly available, with the outcomes of the fourth trial held back for unbiased model evaluation. Challenge participants from around the world trained models and submitted their predictions. Area under the precision-recall curve was the primary metric used for performance assessment. RESULTS: In total, 34 separate teams submitted predictions. Seven models with statistically similar area under precision-recall curves (Bayes factor ≤ 3) outperformed all other models. A postchallenge analysis of risk prediction using these seven models revealed three patient subgroups: high risk, low risk, or discordant risk. Early discontinuation events were two times higher in the high-risk subgroup compared with the low-risk subgroup. Simulation studies demonstrated that use of patient discontinuation prediction models could reduce patient enrollment in clinical trials without the loss of statistical power. CONCLUSION: This work represents a successful collaboration between 34 international teams that leveraged open clinical trial data. Our results demonstrate that routinely collected clinical features can be used to identify patients with mCRPC who are likely to discontinue treatment because of adverse events and establishes a robust benchmark with implications for clinical trial design.

A rat RNA-Seq transcriptomic BodyMap across 11 organs and 4 developmental stages.

The rat has been used extensively as a model for evaluating chemical toxicities and for understanding drug mechanisms. However, its transcriptome across multiple organs, or developmental stages, has not yet been reported. Here we show, as part of the SEQC consortium efforts, a comprehensive rat transcriptomic BodyMap created by performing RNA-Seq on 320 samples from 11 organs of both sexes of juvenile, adolescent, adult and aged Fischer 344 rats. We catalogue the expression profiles of 40,064 genes, 65,167 transcripts, 31,909 alternatively spliced transcript variants and 2,367 non-coding genes/non-coding RNAs (ncRNAs) annotated in AceView. We find that organ-enriched, differentially expressed genes reflect the known organ-specific biological activities. A large number of transcripts show organ-specific, age-dependent or sex-specific differential expression patterns. We create a web-based, open-access rat BodyMap database of expression profiles with crosslinks to other widely used databases, anticipating that it will serve as a primary resource for biomedical research using the rat model.

Comparison of microarrays and RNA-seq for gene expression analyses of dose-response experiments.

Relative to microarrays, RNA-seq has been reported to offer higher precision estimates of transcript abundance, a greater dynamic range, and detection of novel transcripts. However, previous comparisons of the 2 technologies have not covered dose-response experiments that are relevant to toxicology. Male F344 rats were exposed for 13 weeks to 5 doses of bromobenzene, and liver gene expression was measured using both microarrays and RNA-seq. Multiple normalization methods were evaluated for each technology, and gene expression changes were statistically analyzed using both analysis of variance and benchmark dose (BMD). Fold-change values were highly correlated between the 2 technologies, whereas the -log p values showed lower correlation. RNA-seq detected fewer statistically significant genes at lower doses, but more significant genes based on fold change except when a negative binomial transformation was applied. Overlap in genes significant by both p value and fold change was approximately 30%-40%. Random sampling of the RNA-seq data showed an equivalent number of differentially expressed genes compared with microarrays at ~5 million reads. Quantitative RT-PCR of differentially expressed genes uniquely identified by each technology showed a high degree of confirmation when both fold change and p value were considered. The mean dose-response expression of each gene was highly correlated between technologies, whereas estimates of sample variability and gene-based BMD values showed lower correlation. Differences in BMD estimates and statistical significance may be due, in part, to differences in the dynamic range of each technology and the degree to which normalization corrects genes at either end of the scale.

A genomics-based analysis of relative potencies of dioxin-like compounds in primary rat hepatocytes.

Toxic equivalency factors (TEFs) for dioxin-like compounds are largely based on relative potency (REP) values derived from biochemical endpoints such as enzyme activity. As of yet, REPs based on gene expression changes have not been accounted for in the TEF values. In this study, primary rat hepatocytes were treated for 24h with 11 concentrations of 2,3,7,8-tetrachlorodibenzo-p-dioxin, 2,3,4,7,8-pentachlorodibenzofuran (4-PeCDF), or 2,3,7,8-tetrachlorodibenzofuran (TCDF) ranging from 0.00001 to 100 nM. Differential changes in gene expression were analyzed using analysis of variance to assess the relative contributions of concentration, congener, and the interaction between concentration and congener for each gene. A total of 3283 genes showed significant changes with concentration (false discovery rate < .05 and fold-change ± 1.5 in at least 1 concentration for 1 congener). Among these genes, 399 were significant for both concentration and congener effects indicating parallel concentration-response curves with significant differences in potency. Only 8 genes showed a significant concentration and congener interaction term indicating a minority of genes show nonparallel dose-response curves among the 3 congeners. Benchmark dose (BMD) modeling was used to derive BMD values for induced individual genes and signaling pathways. The REP values for 4-PeCDF and TCDF were generally 3- to 5-fold lower than the World Health Organization (WHO) TEF values on both a gene and pathway basis. These findings suggest that the WHO TEF values may possibly overpredict the potency of these polychlorinated dibenzofuran congeners and demonstrate the importance of identifying functional pathways relevant to the toxicological modes of action for establishing pertinent REPs.

Summarizing the incidence of adverse events using volcano plots and time intervals.

BACKGROUND: Adverse event incidence analyses are a critical component for describing the safety profile of any new intervention. The results typically are presented in lengthy summary tables. For therapeutic areas where patients have frequent adverse events, analysis and interpretation are made more difficult by the sheer number and variety of events that occur. Understanding the risk in these instances becomes even more crucial. PURPOSE: We describe a space-saving graphical summary that overcomes the limitations of traditional presentations of adverse events and improves interpretability of the safety profile. METHODS: We present incidence analyses of adverse events graphically using volcano plots to highlight treatment differences. Data from a clinical trial of patients experiencing an aneurysmal subarachnoid hemorrhage are used for illustration. Adjustments for multiplicity are illustrated. RESULTS: Color is used to indicate the treatment with higher incidence; bubble size represents the total number of events that occur in the treatment arms combined. Adjustments for multiple comparisons are displayed in a manner to indicate clearly those events for which the difference between treatment arms is statistically significant. Furthermore, adverse events can be displayed by time intervals, with multiple volcano plots or animation to appreciate changes in adverse event risk over time. Such presentations can emphasize early differences across treatments that may resolve later or highlight events for which treatment differences may become more substantial with longer follow-up. LIMITATIONS: Treatment arms are compared in a pairwise fashion. CONCLUSIONS: Volcano plots are space-saving tools that emphasize important differences between the adverse event profiles of two treatment arms. They can incorporate multiplicity adjustments in a manner that is straightforward to interpret and, by using time intervals, can illustrate how adverse event risk changes over the course of a clinical trial.

Principal variance component analysis of crop composition data: a case study on herbicide-tolerant cotton.

Compositional studies on genetically modified (GM) and non-GM crops have consistently demonstrated that their respective levels of key nutrients and antinutrients are remarkably similar and that other factors such as germplasm and environment contribute more to compositional variability than transgenic breeding. We propose that graphical and statistical approaches that can provide meaningful evaluations of the relative impact of different factors to compositional variability may offer advantages over traditional frequentist testing. A case study on the novel application of principal variance component analysis (PVCA) in a compositional assessment of herbicide-tolerant GM cotton is presented. Results of the traditional analysis of variance approach confirmed the compositional equivalence of the GM and non-GM cotton. The multivariate approach of PVCA provided further information on the impact of location and germplasm on compositional variability relative to GM.

Relative impact of incorporating pharmacokinetics on predicting in vivo hazard and mode of action from high-throughput in vitro toxicity assays.

The use of high-throughput in vitro assays has been proposed to play a significant role in the future of toxicity testing. In this study, rat hepatic metabolic clearance and plasma protein binding were measured for 59 ToxCast phase I chemicals. Computational in vitro-to-in vivo extrapolation was used to estimate the daily dose in a rat, called the oral equivalent dose, which would result in steady-state in vivo blood concentrations equivalent to the AC 50 or lowest effective concentration (LEC) across more than 600 ToxCast phase I in vitro assays. Statistical classification analysis was performed using either oral equivalent doses or unadjusted AC 50 /LEC values for the in vitro assays to predict the in vivo effects of the 59 chemicals. Adjusting the in vitro assays for pharmacokinetics did not improve the ability to predict in vivo effects as either a discrete (yes or no) response or a low effect level (LEL) on a continuous dose scale. Interestingly, a comparison of the in vitro assay with the lowest oral equivalent dose with the in vivo endpoint with the lowest LEL suggested that the lowest oral equivalent dose may provide a conservative estimate of the point of departure for a chemical in a dose-response assessment. Furthermore, comparing the oral equivalent doses for the in vitro assays with the in vivo dose range that resulted in adverse effects identified more coincident in vitro assays across chemicals than expected by chance, suggesting that the approach may also be used to identify potential molecular initiating events leading to adversity.

Anti-Oligomannose Antibodies as Potential Serum Biomarkers of Aggressive Prostate Cancer.

[Table: see text] This study bridges a carbohydrate microarray discovery and a large-scale serological validation of anti-oligomannose antibodies as novel serum biomarkers of aggressive prostate cancer (PCa). Experimentally, a Man9-cluster-specific enzyme-linked immunosorbent assay was established to enable sensitive detection of anti-Man9 antibodies in human sera. A large-cohort of men with PCa or benign prostatic hyperplasia (BPH) whose sera were banked at Stanford University was characterized using this assay. Subjects included patients with 100% Gleason grade 3 cancer (n = 84), with Gleason grades 4 and/or 5 cancer (n = 204), and BPH controls (n = 135). Radical prostatectomy Gleason grades and biochemical (PSA) recurrence served as key parameters for serum biomarker evaluation. It was found that IgGMan9 and IgMMan9 were widely present in the sera of men with BPH, as well as those with cancer. However, these antibody reactivities were significantly increased in the subjects with the largest volumes of high grade cancer. Detection of serum IgGMan9 and IgMMan9 significantly predicted the clinical outcome of PCa post-radical prostatectomy. Given these results, we suggest that IgGMan9 and IgMMan9 are novel serum biomarkers for monitoring aggressive progression of PCa. The potential of oligomannosyl antigens as targets for PCa subtyping and targeted immunotherapy is yet to be explored.

Technical reproducibility of genotyping SNP arrays used in genome-wide association studies.

During the last several years, high-density genotyping SNP arrays have facilitated genome-wide association studies (GWAS) that successfully identified common genetic variants associated with a variety of phenotypes. However, each of the identified genetic variants only explains a very small fraction of the underlying genetic contribution to the studied phenotypic trait. Moreover, discordance observed in results between independent GWAS indicates the potential for Type I and II errors. High reliability of genotyping technology is needed to have confidence in using SNP data and interpreting GWAS results. Therefore, reproducibility of two widely genotyping technology platforms from Affymetrix and Illumina was assessed by analyzing four technical replicates from each of the six individuals in five laboratories. Genotype concordance of 99.40% to 99.87% within a laboratory for the sample platform, 98.59% to 99.86% across laboratories for the same platform, and 98.80% across genotyping platforms was observed. Moreover, arrays with low quality data were detected when comparing genotyping data from technical replicates, but they could not be detected according to venders' quality control (QC) suggestions. Our results demonstrated the technical reliability of currently available genotyping platforms but also indicated the importance of incorporating some technical replicates for genotyping QC in order to improve the reliability of GWAS results. The impact of discordant genotypes on association analysis results was simulated and could explain, at least in part, the irreproducibility of some GWAS findings when the effect size (i.e. the odds ratio) and the minor allele frequencies are low.

A comprehensive statistical analysis of predicting in vivo hazard using high-throughput in vitro screening.

Over the past 5 years, increased attention has been focused on using high-throughput in vitro screening for identifying chemical hazards and prioritizing chemicals for additional in vivo testing. The U.S. Environmental Protection Agency's ToxCast program has generated a significant amount of high-throughput screening data allowing a broad-based assessment of the utility of these assays for predicting in vivo responses. In this study, a comprehensive cross-validation model comparison was performed to evaluate the predictive performance of the more than 600 in vitro assays from the ToxCast phase I screening effort across 60 in vivo endpoints using 84 different statistical classification methods. The predictive performance of the in vitro assays was compared and combined with that from chemical structure descriptors. With the exception of chronic in vivo cholinesterase inhibition, the overall predictive power of both the in vitro assays and the chemical descriptors was relatively low. The predictive power of the in vitro assays was not significantly different from that of the chemical descriptors and aggregating the assays based on genes reduced predictive performance. Prefiltering the in vitro assay data outside the cross-validation loop, as done in some previous studies, significantly biased estimates of model performance. The results suggest that the current ToxCast phase I assays and chemicals have limited applicability for predicting in vivo chemical hazards using standard statistical classification methods. However, if viewed as a survey of potential molecular initiating events and interpreted as risk factors for toxicity, the assays may still be useful for chemical prioritization.

Kernel density weighted loess normalization improves the performance of detection within asymmetrical data.

BACKGROUND: Normalization of gene expression data has been studied for many years and various strategies have been formulated to deal with various types of data. Most normalization algorithms rely on the assumption that the number of up-regulated genes and the number of down-regulated genes are roughly the same. However, the well-known Golden Spike experiment presents a unique situation in which differentially regulated genes are biased toward one direction, thereby challenging the conclusions of previous bench mark studies. RESULTS: This study proposes two novel approaches, KDL and KDQ, based on kernel density estimation to improve upon the basic idea of invariant set selection. The key concept is to provide various importance scores to data points on the MA plot according to their proximity to the cluster of the null genes under the assumption that null genes are more densely distributed than those that are differentially regulated. The comparison is demonstrated in the Golden Spike experiment as well as with simulation data using the ROC curves and compression rates. KDL and KDQ in combination with GCRMA provided the best performance among all approaches. CONCLUSIONS: This study determined that methods based on invariant sets are better able to resolve the problem of asymmetry. Normalization, either before or after expression summary for probesets, improves performance to a similar degree.

Gene set analysis of genome-wide association studies: methodological issues and perspectives.

Recent studies have demonstrated that gene set analysis, which tests disease association with genetic variants in a group of functionally related genes, is a promising approach for analyzing and interpreting genome-wide association studies (GWAS) data. These approaches aim to increase power by combining association signals from multiple genes in the same gene set. In addition, gene set analysis can also shed more light on the biological processes underlying complex diseases. However, current approaches for gene set analysis are still in an early stage of development in that analysis results are often prone to sources of bias, including gene set size and gene length, linkage disequilibrium patterns and the presence of overlapping genes. In this paper, we provide an in-depth review of the gene set analysis procedures, along with parameter choices and the particular methodology challenges at each stage. In addition to providing a survey of recently developed tools, we also classify the analysis methods into larger categories and discuss their strengths and limitations. In the last section, we outline several important areas for improving the analytical strategies in gene set analysis.