Lessons from genetics: interpreting complex phenotypes in RNAi screens

Sustainable Utilization of Machine-Vision-Technique-Based Algorithm in Objective Evaluation of Confocal Microscope Images

Confocal microscope imaging has become popular in biotechnology labs. Confocal imaging technology utilizes fluorescence optics, where laser light is focused onto a specific spot at a defined depth in the sample. A considerable number of images are produced regularly during the process of research. These images require methods of unbiased quantification to have meaningful analyses. Increasing efforts to tie reimbursement to outcomes will likely increase the need for objective data in analyzing confocal microscope images in the coming years. Utilizing visual quantification methods to quantify confocal images with naked human eyes is an essential but often underreported outcome measure due to the time required for manual counting and estimation. The current method (visual quantification methods) of image quantification is time-consuming and cumbersome, and manual measurement is imprecise because of the natural differences among human eyes’ abilities. Subsequently, objective outcome evaluation can obviate the drawbacks of the current methods and facilitate recording for documenting function and research purposes. To achieve a fast and valuable objective estimation of fluorescence in each image, an algorithm was designed based on machine vision techniques to extract the targeted objects in images that resulted from confocal images and then estimate the covered area to produce a percentage value similar to the outcome of the current method and is predicted to contribute to sustainable biotechnology image analyses by reducing time and labor consumption. The results show strong evidence that t-designed objective algorithm evaluations can replace the current method of manual and visual quantification methods to the extent that the Intraclass Correlation Coefficient (ICC) is 0.9.

A critical role of PRDM14 in human primordial germ cell fate revealed by inducible degrons

PRDM14 is a crucial regulator of mouse primordial germ cells (mPGCs), epigenetic reprogramming and pluripotency, but its role in the evolutionarily divergent regulatory network of human PGCs (hPGCs) remains unclear. Besides, a previous knockdown study indicated that PRDM14 might be dispensable for human germ cell fate. Here, we decided to use inducible degrons for a more rapid and comprehensive PRDM14 depletion. We show that PRDM14 loss results in significantly reduced specification efficiency and an aberrant transcriptome of hPGC-like cells (hPGCLCs) obtained in vitro from human embryonic stem cells (hESCs). Chromatin immunoprecipitation and transcriptomic analyses suggest that PRDM14 cooperates with TFAP2C and BLIMP1 to upregulate germ cell and pluripotency genes, while repressing WNT signalling and somatic markers. Notably, PRDM14 targets are not conserved between mouse and human, emphasising the divergent molecular mechanisms of PGC specification. The effectiveness of degrons for acute protein depletion is widely applicable in various developmental contexts.

KCML: a machine-learning framework for inference of multi-scale gene functions from genetic perturbation screens

Characterising context-dependent gene functions is crucial for understanding the genetic bases of health and disease. To date, inference of gene functions from large-scale genetic perturbation screens is based on ad hoc analysis pipelines involving unsupervised clustering and functional enrichment. We present Knowledge- and Context-driven Machine Learning (KCML), a framework that systematically predicts multiple context-specific functions for a given gene based on the similarity of its perturbation phenotype to those with known function. As a proof of concept, we test KCML on three datasets describing phenotypes at the molecular, cellular and population levels and show that it outperforms traditional analysis pipelines. In particular, KCML identified an abnormal multicellular organisation phenotype associated with the depletion of olfactory receptors, and TGFβ and WNT signalling genes in colorectal cancer cells. We validate these predictions in colorectal cancer patients and show that olfactory receptors expression is predictive of worse patient outcomes. These results highlight KCML as a systematic framework for discovering novel scale-crossing and context-dependent gene functions. KCML is highly generalisable and applicable to various large-scale genetic perturbation screens.

Conditional Degrons for Controlling Protein Expression at the Protein Level

The conditional depletion of a protein of interest (POI) is useful not only for loss-of-function studies, but also for the modulation of biological pathways. Technologies that work at the level of DNA, mRNA, and protein are available for temporal protein depletion. Compared with technologies targeting the pretranslation steps, direct protein depletion (or protein knockdown approaches) is advantageous in terms of specificity, reversibility, and time required for depletion, which can be achieved by fusing a POI with a protein domain called a degron that induces rapid proteolysis of the fusion protein. Conditional degrons can be activated or inhibited by temperature, small molecules, light, or the expression of another protein. The conditional degron-based technologies currently available are described and discussed.

Deep phenotyping unveils hidden traits and genetic relations in subtle mutants

Discovering mechanistic insights from phenotypic information is critical for the understanding of biological processes. For model organisms, unlike in cell culture, this is currently bottlenecked by the non-quantitative nature and perceptive biases of human observations, and the limited number of reporters that can be simultaneously incorporated in live animals. An additional challenge is that isogenic populations exhibit significant phenotypic heterogeneity. These difficulties limit genetic approaches to many biological questions. To overcome these bottlenecks, we developed tools to extract complex phenotypic traits from images of fluorescently labelled subcellular landmarks, using C. elegans synapses as a test case. By population-wide comparisons, we identified subtle but relevant differences inaccessible to subjective conceptualization. Furthermore, the models generated testable hypotheses of how individual alleles relate to known mechanisms or belong to new pathways. We show that our model not only recapitulates current knowledge in synaptic patterning but also identifies novel alleles overlooked by traditional methods.

Experimenter scoring of cellular imaging data can be biased. This study describes an automated and unbiased multidimensional phenotyping method that relies on machine learning and complex feature computation of imaging data, and identifies weak alleles affecting synapse morphology in live C. elegans.

Integrating evo-devo with ecology for a better understanding of phenotypic evolution

Evolutionary developmental biology (evo-devo) has provided invaluable contributions to our understanding of the mechanistic relationship between genotypic and phenotypic change. Similarly, evolutionary ecology has greatly advanced our understanding of the relationship between the phenotype and the environment. To fully understand the evolution of organismal diversity, a thorough integration of these two fields is required. This integration remains highly challenging because model systems offering a rich ecological and evolutionary background, together with the availability of developmental genetic tools and genomic resources, are scarce. In this review, we introduce the semi-aquatic bugs (Gerromorpha, Heteroptera) as original models well suited to study why and how organisms diversify. The Gerromorpha invaded water surfaces over 200 mya and diversified into a range of remarkable new forms within this new ecological habitat. We summarize the biology and evolutionary history of this group of insects and highlight a set of characters associated with the habitat change and the diversification that followed. We further discuss the morphological, behavioral, molecular and genomic tools available that together make semi-aquatic bugs a prime model for integration across disciplines. We present case studies showing how the implementation and combination of these approaches can advance our understanding of how the interaction between genotypes, phenotypes and the environment drives the evolution of distinct morphologies. Finally, we explain how the same set of experimental designs can be applied in other systems to address similar biological questions.

Fluorescence microscopy-based RNA interference screening

Using RNAi interference (RNAi), it is possible to study the effect of specific gene knockdowns in mammalian cells. In this protocol we present the automated preparation of "ready to transfect" multiwell plates and cell arrays, on which cells can be grown which are then reversely transfected with one type of siRNA in every individual well or spot. Additionally, different microscope types for screening approaches are compared and considerations about the information workflow are made.

Exploiting cell-to-cell variability to detect cellular perturbations

Any single-cell-resolved measurement generates a population distribution of phenotypes, characterized by a mean, a variance, and a shape. Here we show that changes in the shape of a phenotypic distribution can signal perturbations to cellular processes, providing a way to screen for underlying molecular machinery. We analyzed images of a Drosophila S2R+ cell line perturbed by RNA interference, and tracked 27 single-cell features which report on endocytic activity, and cell and nuclear morphology. In replicate measurements feature distributions had erratic means and variances, but reproducible shapes; RNAi down-regulation reliably induced shape deviations in at least one feature for 1072 out of 7131 genes surveyed, as revealed by a Kolmogorov-Smirnov-like statistic. We were able to use these shape deviations to identify a spectrum of genes that influenced cell morphology, nuclear morphology, and multiple pathways of endocytosis. By preserving single-cell data, our method was even able to detect effects invisible to a population-averaged analysis. These results demonstrate that cell-to-cell variability contains accessible and useful biological information, which can be exploited in existing cell-based assays.

Matrix stiffness affects endocytic uptake of MK2-inhibitor peptides

In this study, the role of substrate stiffness on the endocytic uptake of a cell-penetrating peptide was investigated. The cell-penetrating peptide, an inhibitor of mitogen-activated protein kinase activated protein kinase II (MK2), enters a primary mesothelial cell line predominantly through caveolae. Using tissue culture polystyrene and polyacrylamide gels of varying stiffness for cell culture, and flow cytometry quantification and enzyme-linked immunoassays (ELISA) for uptake assays, we showed that the amount of uptake of the peptide is increased on soft substrates. Further, peptide uptake per cell increased at lower cell density. The improved uptake seen on soft substrates in vitro better correlates with in vivo functional studies where 10–100 µM concentrations of the MK2 inhibitor cell penetrating peptide demonstrated functional activity in several disease models. Additional characterization showed actin polymerization did not affect uptake, while microtubule polymerization had a profound effect on uptake. This work demonstrates that cell culture substrate stiffness can play a role in endocytic uptake, and may be an important consideration to improve correlations between in vitro and in vivo drug efficacy.

Unsupervised automated high throughput phenotyping of RNAi time-lapse movies

Background

Gene perturbation experiments in combination with fluorescence time-lapse cell imaging are a powerful tool in reverse genetics. High content applications require tools for the automated processing of the large amounts of data. These tools include in general several image processing steps, the extraction of morphological descriptors, and the grouping of cells into phenotype classes according to their descriptors. This phenotyping can be applied in a supervised or an unsupervised manner. Unsupervised methods are suitable for the discovery of formerly unknown phenotypes, which are expected to occur in high-throughput RNAi time-lapse screens.

Results

We developed an unsupervised phenotyping approach based on Hidden Markov Models (HMMs) with multivariate Gaussian emissions for the detection of knockdown-specific phenotypes in RNAi time-lapse movies. The automated detection of abnormal cell morphologies allows us to assign a phenotypic fingerprint to each gene knockdown. By applying our method to the Mitocheck database, we show that a phenotypic fingerprint is indicative of a gene’s function.

Conclusion

Our fully unsupervised HMM-based phenotyping is able to automatically identify cell morphologies that are specific for a certain knockdown. Beyond the identification of genes whose knockdown affects cell morphology, phenotypic fingerprints can be used to find modules of functionally related genes.

A Novel Organelle Map Framework for High-Content Cell Morphology Analysis in High Throughput

A screening procedure was developed that takes advantage of the cellular normalization by micropatterning and a novel quantitative organelle mapping approach that allows unbiased and automated cell morphology comparison using black-box statistical testing. Micropatterns of extracellular matrix proteins force cells to adopt a reproducible shape and distribution of intracellular compartments avoiding strong cell-to-cell variation that is a major limitation of classical culture conditions. To detect changes in cell morphology induced by compound treatment, fluorescently labeled intracellular structures from several tens of micropatterned cells were transformed into probabilistic density maps. Then, the similarity or difference between two given density maps was quantified using statistical testing that evaluates differences directly from the data without additional analysis or any subjective decision. The versatility of this organelle mapping approach for different magnifications and its performance for different cell shapes has been assessed. Density-based analysis detected changes in cell morphology due to compound treatment in a small-scale proof-of-principle screen demonstrating its compatibility with high-throughput screening. This novel tool for high-content and high-throughput cellular phenotyping can potentially be used for a wide range of applications from drug screening to careful characterization of cellular processes.

A divide and conquer approach for construction of large-scale signaling networks from PPI and RNAi data using linear programming

Inference of topology of signaling networks from perturbation experiments is a challenging problem. Recently, the inference problem has been formulated as a reference network editing problem and it has been shown that finding the minimum number of edit operations on a reference network to comply with perturbation experiments is an NP-complete problem. In this paper, we propose an integer linear optimization (ILP) model for reconstruction of signaling networks from RNAi data and a reference network. The ILP model guarantees the optimal solution; however, is practical only for small signaling networks of size 10-15 genes due to computational complexity. To scale for large signaling networks, we propose a divide and conquer-based heuristic, in which a given reference network is divided into smaller subnetworks that are solved separately and the solutions are merged together to form the solution for the large network. We validate our proposed approach on real and synthetic data sets, and comparison with the state of the art shows that our proposed approach is able to scale better for large networks while attaining similar or better biological accuracy.

A Potential of microRNAs for High-Content Screening

In the last years miRNAs have increasingly been recognised as potent posttranscriptional regulators of gene expression. Possibly, miRNAs exert their action on virtually any biological process by simultaneous regulation of numerous genes. The importance of miRNA-based regulation in health and disease has inspired research to investigate diverse aspects of miRNA origin, biogenesis, and function. Despite the recent rapid accumulation of experimental data, and the emergence of functional models, the complexity of miRNA-based regulation is still far from being well understood. In particular, we lack comprehensive knowledge as to which cellular processes are regulated by which miRNAs, and, furthermore, how temporal and spatial interactions of miRNAs to their targets occur. Results from large-scale functional analyses have immense potential to address these questions. In this review, we discuss the latest progress in application of high-content and high-throughput functional analysis for the systematic elucidation of the biological roles of miRNAs.

Tales from an academic RNAi screening facility; FAQs

RNAi technology is now a well-established and widely employed research technique that has been adopted by many researchers for use in large-scale screening campaigns. Here, we offer our experience of genome-wide siRNA screening from the perspective of a facility providing screening as a service to a wide range of researchers with diverse interests and approaches. We have experienced the emotional rollercoaster of screening from the exuberant early promise of a screen, the messy reality of the data through to the recognition of screen data as a potential information goldmine. Here, we use some of the questions we most frequently encounter to highlight the initial concerns of many researchers embarking on a siRNA screen and conclude that an informed view of what can be reasonably expected from a screen is essential to the most effective implementation of the technology. Along the way, we suggest that for this area of research at least, either centralization of the resources or close and open collaboration between interested parties offers distinct advantages.

Kinase/phosphatase overexpression reveals pathways regulating hippocampal neuron morphology

Kinases and phosphatases that regulate neurite number versus branching versus extension are weakly correlated.

The kinase family that most strongly enhances neurite growth is a family of non-protein kinases; sugar kinases related to NADK.

Pathway analysis revealed that genes in several cancer pathways were highly active in enhancing neurite growth.

In neural development, neuronal precursors differentiate, migrate, extend long axons and dendrites, and finally establish connections with their targets. Clinical conditions such as spinal cord injury, traumatic brain injury, stroke, multiple sclerosis, Parkinson's disease, Huntington's disease, and Alzheimer's disease are often associated with a loss of axon and/or dendrite connectivity and treatment strategies would be enhanced by new therapies targeting cell intrinsic mechanisms of axon elongation and regeneration.

Phosphorylation controls most cellular processes, including the cell cycle, proliferation, metabolism, and apoptosis. Neuronal differentiation, including axon formation and elongation, is also regulated by a wide range of kinases and phosphatases. For example, the non-receptor tyrosine kinase Src is required for cell adhesion molecule-dependent neurite outgrowth. In addition to individual kinases and phosphatases, signaling pathways like the MAPK, growth factor signaling, PIP3, cytoskeletal, and calcium-dependent pathways have been shown to impinge on or control neuronal process development. Recent results have implicated GSK3 and PTEN as therapeutically relevant targets in axonal regeneration after injury. However, these and other experiments have studied only a small fraction of the total kinases and phosphatases in the genome. Because of recent advances in genomic knowledge, large-scale cDNA production, and high-throughput phenotypic analysis, it is now possible to take a more comprehensive approach to understanding the functions of kinases and phosphatases in neurons.

We performed a large, unbiased set of experiments to answer the question ‘what effect does the overexpression of genes encoding kinases, phosphatases, and related proteins have on neuronal morphology?' We used ‘high-content analysis' to obtain detailed results about the specific phenotypes of neurons. We studied embryonic rat hippocampal neurons because of their stereotypical development in vitro (Dotti et al, 1988) and their widespread use in studies of neuronal differentiation and signaling. We transfected over 700 clones encoding kinases and phosphatases into hippocampal neurons and analyzed the resulting changes in neuronal morphology.

Many known genes, including PP1a, ERK1, ErbB2, atypical PKC, Calcineurin, CaMK2, IGF1R, FGFR, GSK3, and PIK3 were observed to have significant effects on neurite outgrowth in our system, consistent with earlier findings in the literature.

We obtained quantitative data for many cellular and neuronal morphological parameters from each neuron imaged. These included nuclear morphology (nuclear area and Hoechst dye intensity), soma morphology (tubulin intensity, area, and shape), and numerous parameters of neurite morphology (e.g. tubulin intensity along the neurites, number of primary neurites, neurite length, number of branches, distance from the cell body to the branches, number of crossing points, width and area of the neurites, and longest neurite; Supplementary Figure 1). Other parameters were reported on a ‘per well' basis, including the percentage of transfected neurons in a condition, as well as the percentage of neurons initiating neurite growth. Data for each treatment were normalized to a control (pSport CAT) within the same experiment, then aggregated across replicate experiments.

Correlations among the 19 normalized parameters were analyzed for neurons transfected with all kinase and phosphatase clones (Figure 2). On the basis of this analysis, the primary variables that define the neurite morphology are primary neurite count, neurite average length, and average branches. Interestingly, primary neurite count was not well correlated with neurite length or branching. The Pearson correlation coefficient (r²) between the number of primary neurites and the average length of the neurites was 0.3, and between the number of primary neurites and average branching was 0.2. In contrast, the correlation coefficient of average branching with neurite average length was 0.7. The most likely explanation is that signaling mechanisms underlying the neurite number determination are different than those controlling length/branching of the neurites.

Related proteins are often involved in similar neuronal functions. For example, families of receptor protein tyrosine phosphatases are involved in motor axon extension and guidance in both Drosophila and in vertebrates, and a large family of Eph receptor tyrosine kinases regulates guidance of retinotectal projections, motor axons, and axons in the corpus callosum. We therefore asked whether families of related genes produced similar phenotypes when overexpressed in hippocampal neurons. Our set of genes covered 40% of the known protein kinases, and many of the non-protein kinases and phosphatases.

Gene families commonly exhibit redundant function. Redundant gene function has often been identified when two or more knockouts are required to produce a phenotype. Our technique allowed us to measure whether different members of gene families had similar (potentially redundant) or distinct effects on neuronal phenotype.

To determine whether groups of related genes affect neuronal morphology in similar ways, we used sequence alignment information to construct gene clusters (Figure 6). Genes were clustered at nine different thresholds of similarity (called ‘tiers'). The functional effect for a particular parameter was then averaged within each cluster of a given tier, and statistics were performed to determine the significance of the effect. We analyzed the results for three key neurite parameters (average neurite length, primary neurite count, and average branching). Genes that perturbed each of these phenotypes are grouped in Figure 6. Eight families, most with only a few genes, produced significant changes for one or two parameters. A diverse family of non-protein kinases had a positive effect on neurite outgrowth in three of the four parameters analyzed. This family of kinases consisted of a variety of enzymes, mostly sugar and lipid kinases. A similar analysis was performed using pathway cluster analysis with pathways from the KEGG database, rather than sequence homology. Interestingly, pathways involved in cancer cell proliferation potentiated neurite extension and branching.

Our studies have identified a large number of kinases and phosphatases, as well as structurally and functionally defined families of these proteins, that affect neuronal process formation in specific ways. We have provided an analytical methodology and new tools to analyze functional data, and have implicated genes with novel functions in neuronal development. Our studies are an important step towards the goal of a molecular description of the intrinsic control of axodendritic growth.

Development and regeneration of the nervous system requires the precise formation of axons and dendrites. Kinases and phosphatases are pervasive regulators of cellular function and have been implicated in controlling axodendritic development and regeneration. We undertook a gain-of-function analysis to determine the functions of kinases and phosphatases in the regulation of neuron morphology. Over 300 kinases and 124 esterases and phosphatases were studied by high-content analysis of rat hippocampal neurons. Proteins previously implicated in neurite growth, such as ERK1, GSK3, EphA8, FGFR, PI3K, PKC, p38, and PP1a, were confirmed to have effects in our functional assays. We also identified novel positive and negative neurite growth regulators. These include neuronal-developmentally regulated kinases such as the activin receptor, interferon regulatory factor 6 (IRF6) and neural leucine-rich repeat 1 (LRRN1). The protein kinase N2 (PKN2) and choline kinase α (CHKA) kinases, and the phosphatases PPEF2 and SMPD1, have little or no established functions in neuronal function, but were sufficient to promote neurite growth. In addition, pathway analysis revealed that members of signaling pathways involved in cancer progression and axis formation enhanced neurite outgrowth, whereas cytokine-related pathways significantly inhibited neurite formation.

Evaluation of methods for detection of fluorescence labeled subcellular objects in microscope images

Background

Several algorithms have been proposed for detecting fluorescently labeled subcellular objects in microscope images. Many of these algorithms have been designed for specific tasks and validated with limited image data. But despite the potential of using extensive comparisons between algorithms to provide useful information to guide method selection and thus more accurate results, relatively few studies have been performed.

Results

To better understand algorithm performance under different conditions, we have carried out a comparative study including eleven spot detection or segmentation algorithms from various application fields. We used microscope images from well plate experiments with a human osteosarcoma cell line and frames from image stacks of yeast cells in different focal planes. These experimentally derived images permit a comparison of method performance in realistic situations where the number of objects varies within image set. We also used simulated microscope images in order to compare the methods and validate them against a ground truth reference result. Our study finds major differences in the performance of different algorithms, in terms of both object counts and segmentation accuracies.

Conclusions

These results suggest that the selection of detection algorithms for image based screens should be done carefully and take into account different conditions, such as the possibility of acquiring empty images or images with very few spots. Our inclusion of methods that have not been used before in this context broadens the set of available detection methods and compares them against the current state-of-the-art methods for subcellular particle detection.

Simplistic pathways or complex networks?

Signaling events are frequently described in textbooks as linear cascades. However, in reality, input cues are processed by dynamic and context-specific networks, which are assembled from numerous signaling molecules. Diseases, such as cancer, are typically associated with multiple genomic alterations that likely change the structure and dynamics of cellular signaling networks. To assess the impact of such genomic alterations on the structure of signaling networks and on the ability of cells to accurately translate environmental cues into phenotypic changes, we argue studies must be conducted on a network level. Advances in technologies and computational approaches for data integration have permitted network studies of signaling events in both cancer and normal cells. Here we will review recent advances and how they have impacted our view on signaling networks with a specific angle on signal processing in cancer.

Population context determines cell-to-cell variability in endocytosis and virus infection

Single-cell heterogeneity in cell populations arises from a combination of intrinsic and extrinsic factors. This heterogeneity has been measured for gene transcription, phosphorylation, cell morphology and drug perturbations, and used to explain various aspects of cellular physiology. In all cases, however, the causes of heterogeneity were not studied. Here we analyse, for the first time, the heterogeneous patterns of related cellular activities, namely virus infection, endocytosis and membrane lipid composition in adherent human cells. We reveal correlations with specific cellular states that are defined by the population context of a cell, and we derive probabilistic models that can explain and predict most cellular heterogeneity of these activities, solely on the basis of each cell's population context. We find that accounting for population-determined heterogeneity is essential for interpreting differences between the activity levels of cell populations. Finally, we reveal that synergy between two molecular components, focal adhesion kinase and the sphingolipid GM1, enhances the population-determined pattern of simian virus 40 (SV40) infection. Our findings provide an explanation for the origin of heterogeneity patterns of cellular activities in adherent cell populations.

Knowledge based identification of essential signaling from genome-scale siRNA experiments

Background

A systems biology interpretation of genome-scale RNA interference (RNAi) experiments is complicated by scope, experimental variability and network signaling robustness. Over representation approaches (ORA), such as the Hypergeometric or z-score, are an established statistical framework used to associate RNA interference effectors to biologically annotated gene sets or pathways. These methods, however, do not directly take advantage of our growing understanding of the interactome. Furthermore, these methods can miss partial pathway activation and may be biased by protein complexes. Here we present a novel ORA, protein interaction permutation analysis (PIPA), that takes advantage of canonical pathways and established protein interactions to identify pathways enriched for protein interactions connecting RNAi hits.

Results

We use PIPA to analyze genome-scale siRNA cell growth screens performed in HeLa and TOV cell lines. First we show that interacting gene pair siRNA hits are more reproducible than single gene hits. Using protein interactions, PIPA identifies enriched pathways not found using the standard Hypergeometric analysis including the FAK cytoskeletal remodeling pathway. Different branches of the FAK pathway are distinctly essential in HeLa versus TOV cell lines while other portions are uneffected by siRNA perturbations. Enriched hits belong to protein interactions associated with cell cycle regulation, anti-apoptosis, and signal transduction.

Conclusion

PIPA provides an analytical framework to interpret siRNA screen data by merging biologically annotated gene sets with the human interactome. As a result we identify pathways and signaling hypotheses that are statistically enriched to effect cell growth in human cell lines. This method provides a complementary approach to standard gene set enrichment that utilizes the additional knowledge of specific interactions within biological gene sets.