HISTONE DEACETYLASE 9 promotes hypocotyl-specific auxin response under shade.

Vegetative shade causes an array of morphological changes in plants called shade avoidance syndrome, which includes hypocotyl and petiole elongation, leaf hyponasty, reduced leaf growth, early flowering and rapid senescence. Here, we show that loss-of-function mutations in HISTONE DEACETYLASE 9 (HDA9) attenuated the shade-induced hypocotyl elongation in Arabidopsis. However, the hda9 cotyledons and petioles under shade were not significantly different from those in wild-type, suggesting a specific function of HDA9 in hypocotyl elongation in response to shade. HDA9 expression levels were stable under shade and its protein was ubiquitously detected in cotyledon, hypocotyl and root. Organ-specific transcriptome analysis unraveled that shade induced a set of auxin-responsive genes, such as SMALL AUXIN UPREGULATED RNAs (SAURs) and AUXIN/INDOLE-3-ACETIC ACIDs (AUX/IAAs) and their induction was impaired in hda9-1 hypocotyls. In addition, HDA9 binding to loci of SAUR15/65, IAA5/6/19 and ACS4 was increased under shade. The genetic and organ-specific gene expression analyses further revealed that HDA9 may cooperate with PHYTOCHROME-INTERACTING FACTOR 4/7 in the regulation of shade-induced hypocotyl elongation. Furthermore, HDA9 and PIF7 proteins were found to interact together and thus it is suggested that PIF7 may recruit HDA9 to regulate the shade/auxin responsive genes in response to shade. Overall, our study unravels that HDA9 can work as one component of a hypocotyl-specific transcriptional regulatory machinery that activates the auxin response at the hypocotyl leading to the elongation of this organ under shade.

Machine learning alternative to systems biology should not solely depend on data.

In recent years, artificial intelligence (AI)/machine learning has emerged as a plausible alternative to systems biology for the elucidation of biological phenomena and in attaining specified design objective in synthetic biology. Although considered highly disruptive with numerous notable successes so far, we seek to bring attention to both the fundamental and practical pitfalls of their usage, especially in illuminating emergent behaviors from chaotic or stochastic systems in biology. Without deliberating on their suitability and the required data qualities and pre-processing approaches beforehand, the research and development community could experience similar 'AI winters' that had plagued other fields. Instead, we anticipate the integration or combination of the two approaches, where appropriate, moving forward.

Integration of external signaling pathways with the core transcriptional network in embryonic stem cells.

Transcription factors (TFs) and their specific interactions with targets are crucial for specifying gene-expression programs. To gain insights into the transcriptional regulatory networks in embryonic stem (ES) cells, we use chromatin immunoprecipitation coupled with ultra-high-throughput DNA sequencing (ChIP-seq) to map the locations of 13 sequence-specific TFs (Nanog, Oct4, STAT3, Smad1, Sox2, Zfx, c-Myc, n-Myc, Klf4, Esrrb, Tcfcp2l1, E2f1, and CTCF) and 2 transcription regulators (p300 and Suz12). These factors are known to play different roles in ES-cell biology as components of the LIF and BMP signaling pathways, self-renewal regulators, and key reprogramming factors. Our study provides insights into the integration of the signaling pathways into the ES-cell-specific transcription circuitries. Intriguingly, we find specific genomic regions extensively targeted by different TFs. Collectively, the comprehensive mapping of TF-binding sites identifies important features of the transcriptional regulatory networks that define ES-cell identity.

Combined multivariate statistical and flux balance analyses uncover media bottlenecks to the growth and productivity of Chinese hamster ovary cell cultures.

Chinese hamster ovary (CHO) cells are widely used for producing recombinant proteins. To enhance their productivity and product quality, media reformulation has been a key strategy, albeit with several technical challenges, due to the myriad of complex molecular mechanisms underlying media effects on culture performance. Thus, it is imperative to characterize metabolic bottlenecks under various media conditions systematically. To do so, we combined partial least square regression (PLS-R) with the flux balance analysis of a genome-scale metabolic model to elucidate the physiological states and metabolic behaviors of human alpha-1 antitrypsin producing CHO-DG44 cells grown in one commercial and another two in-house media under development. At the onset, PLS-R was used to identify metabolite exchanges that were correlated to specific growth and productivity. Then, by comparing metabolic states described by resultant flux distributions under two of the media conditions, we found suboptimal level of four nutrients and two metabolic wastes, which plausibly hindered cellular growth and productivity; mechanistically, lactate and ammonia recycling were modulated by glutamine and asparagine metabolisms in the media conditions, and also by hitherto unsuspected folate and choline supplements. Our work demonstrated how multivariate statistical analysis can be synergistically combined with metabolic modeling to uncover the mechanistic elements underlying differing media performance. It thus paved the way for the systematic identification of nutrient targets for medium reformulation to enhance recombinant protein production in CHO cells.

Comparative Transcriptome Analysis Reveals Coordinated Transcriptional Regulation of Central and Secondary Metabolism in the Trichomes of Cannabis Cultivars.

Cannabis is one of the few plant genera capable of producing cannabinoids, the effects of which are synergized by terpene interactions. As the biosynthesis of both metabolite classes requires the same intracellular feedstocks, this work describes the coordinated regulation of global metabolic pathways that allows for their joint copious production in vivo. To this end, a transcriptomics-based approach to characterize the glandular trichomes of five Cannabis cultivars was pursued. Besides revealing metabolic traits that enhanced and proportionated the supply of critical carbon precursors, in-depth analysis showed significantly increased gene expression of two particular enzymes to meet the huge nicotinamide adenine dinucleotide phosphate (NADPH) demand of secondary metabolite production. Furthermore, it led to a hypothesis that the methyl-d-erythritol 4-phosphate pathway might be utilized more than the mevalonic acid pathway in Cannabis trichomes. While both pathways were found to be activated in a modular and calibrated way that reflected their broad participation in physiological processes, the genes for hexanoate, cannabinoid, and terpene biosynthesis were, in contrast, up-regulated in an en bloc and multi-loci manner due to their specific roles in secondary metabolite production. In addition, three new terpene synthases were characterized based on both in silico and experimental assays. Altogether, the study enhances the current understanding of secondary metabolite production in Cannabis cultivars, which may assist in their characterization and development.

Comparative phenotypic and transcriptomic analyses unravel conserved and distinct mechanisms underlying shade avoidance syndrome in Brassicaceae vegetables.

BACKGROUND: Plants grown under shade are exposed to low red/far-red ratio, thereby triggering an array of altered phenotypes called shade avoidance syndrome (SAS). Shade negatively influences plant growth, leading to a reduction in agricultural productivity. Understanding of SAS is crucial for sustainable agricultural practices, especially for high-density indoor farming. Brassicaceae vegetables are widely consumed around the world and are commonly cultivated in indoor farms. However, our understanding of SAS in Brassicaceae vegetables and their genome-wide transcriptional regulatory networks are still largely unexplored. RESULTS: Shade induced common signs of SAS, including hypocotyl elongation and reduced carotenoids/anthocyanins biosynthesis, in two different Brassicaceae species: Brassica rapa (Choy Sum and Pak Choy) and Brassica oleracea (Kai Lan). Phenotype-assisted transcriptome analysis identified a set of genes induced by shade in these species, many of which were related to auxin biosynthesis and signaling [e.g. YUCCA8 (YUC8), YUC9, and INDOLE-3-ACETIC ACID INDUCIBLE (IAAs)] and other phytohormones signaling pathways including brassinosteroids and ethylene. The genes functioning in plant defense (e.g. MYB29 and JASMONATE-ZIM-DOMAIN PROTEIN 9) as well as in biosynthesis of anthocyanins and glucosinolates were repressed upon shade. Besides, each species also exhibited distinct SAS phenotypes. Shade strongly reduced primary roots and elongated petioles of B. oleracea, Kai Lan. However, these SAS phenotypes were not clearly recognized in B. rapa, Choy Sum and Pak Choy. Some auxin signaling genes (e.g. AUXIN RESPONSE FACTOR 19, IAA10, and IAA20) were specifically induced in B. oleracea, while homologs in B. rapa were not up-regulated under shade. Contrastingly, shade-exposed B. rapa vegetables triggered the ethylene signaling pathway earlier than B. oleracea, Kai Lan. Interestingly, shade induced the transcript levels of LONG HYPOCOTYL IN FAR-RED 1 (HFR1) homolog in only Pak Choy as B. rapa. As HFR1 is a key negative regulator of SAS in Arabidopsis, our finding suggests that Pak Choy HFR1 homolog may also function in conferring higher shade tolerance in this variety. CONCLUSIONS: Our study shows that two Brassicaceae species not only share a conserved SAS mechanism but also exhibit distinct responses to shade, which will provide comprehensive information to develop new shade-tolerant cultivars that are suitable for high-density indoor farms.

Enzyme capacity-based genome scale modelling of CHO cells.

Chinese hamster ovary (CHO) cells are most prevalently used for producing recombinant therapeutics in biomanufacturing. Recently, more rational and systems approaches have been increasingly exploited to identify key metabolic bottlenecks and engineering targets for cell line engineering and process development based on the CHO genome-scale metabolic model which mechanistically characterizes cell culture behaviours. However, it is still challenging to quantify plausible intracellular fluxes and discern metabolic pathway usages considering various clonal traits and bioprocessing conditions. Thus, we newly incorporated enzyme kinetic information into the updated CHO genome-scale model (iCHO2291) and added enzyme capacity constraints within the flux balance analysis framework (ecFBA) to significantly reduce the flux variability in biologically meaningful manner, as such improving the accuracy of intracellular flux prediction. Interestingly, ecFBA could capture the overflow metabolism under the glucose excess condition where the usage of oxidative phosphorylation is limited by the enzyme capacity. In addition, its applicability was successfully demonstrated via a case study where the clone- and media-specific lactate metabolism was deciphered, suggesting that the lactate-pyruvate cycling could be beneficial for CHO cells to efficiently utilize the mitochondrial redox capacity. In summary, iCHO2296 with ecFBA can be used to confidently elucidate cell cultures and effectively identify key engineering targets, thus guiding bioprocess optimization and cell engineering efforts as a part of digital twin model for advanced biomanufacturing in future.

An LC-MS-based lipidomics pre-processing framework underpins rapid hypothesis generation towards CHO systems biotechnology.

INTRODUCTION: Given a raw LC-MS dataset, it is often required to rapidly generate initial hypotheses, in conjunction with other 'omics' datasets, without time-consuming lipid verifications. Furthermore, for meta-analysis of many datasets, it may be impractical to conduct exhaustive confirmatory analyses. In other cases, samples for validation may be difficult to obtain, replicate or maintain. Thus, it is critical that the computational identification of lipids is of appropriate accuracy, coverage, and unbiased by a researcher's experience and prior knowledge. OBJECTIVES: We aim to prescribe a systematic framework for lipid identifications, without usage of their characteristic retention-time by fully exploiting their underlying mass features. RESULTS: Initially, a hybrid technique, for deducing both common and distinctive daughter ions, is used to infer parent lipids from deconvoluted spectra. This is followed by parent confirmation using basic knowledge of their preferred product ions. Using the framework, we could achieve an accuracy of ~ 80% by correctly identified 101 species from 18 classes in Chinese hamster ovary (CHO) cells. The resulting inferences could explain the recombinant-producing capability of CHO-SH87 cells, compared to non-producing CHO-K1 cells. For comparison, a XCMS-based study of the same dataset, guided by a user's ad-hoc knowledge, identified less than 60 species of 12 classes from thousands of possibilities. CONCLUSION: We describe a systematic LC-MS-based framework that identifies lipids for rapid hypothesis generation.

Metabolomics and modelling approaches for systems metabolic engineering.

Metabolic engineering involves the manipulation of microbes to produce desirable compounds through genetic engineering or synthetic biology approaches. Metabolomics involves the quantitation of intracellular and extracellular metabolites, where mass spectrometry and nuclear magnetic resonance based analytical instrumentation are often used. Here, the experimental designs, sample preparations, metabolite quenching and extraction are essential to the quantitative metabolomics workflow. The resultant metabolomics data can then be used with computational modelling approaches, such as kinetic and constraint-based modelling, to better understand underlying mechanisms and bottlenecks in the synthesis of desired compounds, thereby accelerating research through systems metabolic engineering. Constraint-based models, such as genome scale models, have been used successfully to enhance the yield of desired compounds from engineered microbes, however, unlike kinetic or dynamic models, constraint-based models do not incorporate regulatory effects. Nevertheless, the lack of time-series metabolomic data generation has hindered the usefulness of dynamic models till today. In this review, we show that improvements in automation, dynamic real-time analysis and high throughput workflows can drive the generation of more quality data for dynamic models through time-series metabolomics data generation. Spatial metabolomics also has the potential to be used as a complementary approach to conventional metabolomics, as it provides information on the localization of metabolites. However, more effort must be undertaken to identify metabolites from spatial metabolomics data derived through imaging mass spectrometry, where machine learning approaches could prove useful. On the other hand, single-cell metabolomics has also seen rapid growth, where understanding cell-cell heterogeneity can provide more insights into efficient metabolic engineering of microbes. Moving forward, with potential improvements in automation, dynamic real-time analysis, high throughput workflows, and spatial metabolomics, more data can be produced and studied using machine learning algorithms, in conjunction with dynamic models, to generate qualitative and quantitative predictions to advance metabolic engineering efforts.

WEbcoli: an interactive and asynchronous web application for in silico design and analysis of genome-scale E.coli model.

SUMMARY: WEbcoli is a WEb application for in silico designing, analyzing and engineering Escherichia coli metabolism. It is devised and implemented using advanced web technologies, thereby leading to enhanced usability and dynamic web accessibility. As a main feature, the WEbcoli system provides a user-friendly rich web interface, allowing users to virtually design and synthesize mutant strains derived from the genome-scale wild-type E.coli model and to customize pathways of interest through a graph editor. In addition, constraints-based flux analysis can be conducted for quantifying metabolic fluxes and charactering the physiological and metabolic states under various genetic and/or environmental conditions. AVAILABILITY: WEbcoli is freely accessible at http://webcoli.org. CONTACT: cheld@nus.edu.sg.

In silico model-based characterization of metabolic response to harsh sparging stress in fed-batch CHO cell cultures.

Mammalian cell culture platform has been successfully implemented for industrial biopharmaceutical production through the advancements in early stage process development including cell-line engineering, media design and process optimization. However, late stage developments such as scale-up, scale-down and large-scale cell cultivation still face many industrial challenges to acquire comparable process performance between different culture scales. One of them is the sparging strategy which significantly affects productivity, quality and comparability. Currently, it is mainly relying on the empirical records due to the lack of theoretical framework and scarcity of available literatures to elucidate intracellular metabolic features. Therefore, it is highly required to characterize the underlying mechanism of physiological changes and metabolic states upon the aeration stress. To this end, initially we cultivated antibody producing CHO cells under mild and harsh sparging conditions and observed that sparging stress leads to the decreased cell growth rate, viability and productivity. Subsequent in silicomodel-driven flux analysis suggested that sparging stress rewires amino acid metabolism towards the enriched H2O2 turnover rate by up-regulated fluxes of amino acid oxidases. Interestingly, many of these H2O2-generating reactions were closely connected with the production of NADH, NADPH and GSH which are typical reducing equivalents. Thus, we can hypothesize that increased amino acid uptake caused by sparging stress contributes to restore redox homeostasis against oxidative stress. The current model-driven systematic data analysis allows us to quickly define distinct metabolic feature under stress condition by using basic cell cultivation datasets.

Assessment of CASP7 structure predictions for template free targets.

In CASP7, protein structure prediction targets that lacked substantial similarity to a protein in the PDB at the time of assessment were considered to be free modeling targets (FM). We assessed predictions for 14 FM targets as well as four other targets that were deemed to be on the borderline between FM targets and template based modeling targets (TBM/FM). GDT_TS was used as one measure of model quality. Model quality was also assessed by visual inspection. Visual inspection was performed by three independent assessors who were blinded to GDT_TS scores and other quantitative measures of model quality. The best models by visual inspection tended to rank among the top few percent by GDT_TS, but were typically not the highest scoring models. Thus, visual inspection remains an essential component of assessment for FM targets. Overall, group TS020 (Baker) performed best, but success on individual targets was widely distributed among many groups. Among these other groups, TS024 and TS025 (Zhang and Zhang server) performed notably well without exceptionally large computing resources. This should be considered encouraging for future CASPs. There was a sense of progress in template FM relative to CASP6, but we were unable to demonstrate this progress objectively.

Mammalian Systems Biotechnology Reveals Global Cellular Adaptations in a Recombinant CHO Cell Line.

Effective development of host cells for therapeutic protein production is hampered by the poor characterization of cellular transfection. Here, we employed a multi-omics-based systems biotechnology approach to elucidate the genotypic and phenotypic differences between a wild-type and recombinant antibody-producing Chinese hamster ovary (CHO) cell line. At the genomic level, we observed extensive rearrangements in specific targeted loci linked to transgene integration sites. Transcriptional re-wiring of DNA damage repair and cellular metabolism in the antibody producer, via changes in gene copy numbers, was also detected. Subsequent integration of transcriptomic data with a genome-scale metabolic model showed a substantial increase in energy metabolism in the antibody producer. Metabolomics, lipidomics, and glycomics analyses revealed an elevation in long-chain lipid species, potentially associated with protein transport and secretion requirements, and a surprising stability of N-glycosylation profiles between both cell lines. Overall, the proposed knowledge-based systems biotechnology framework can further accelerate mammalian cell-line engineering in a targeted manner.

Genome-Wide Transcriptome and Binding Sites Analyses Identify Early FOX Expressions for Enhancing Cardiomyogenesis Efficiency of hESC Cultures.

The differentiation efficiency of human embryonic stem cells (hESCs) into heart muscle cells (cardiomyocytes) is highly sensitive to culture conditions. To elucidate the regulatory mechanisms involved, we investigated hESCs grown on three distinct culture platforms: feeder-free Matrigel, mouse embryonic fibroblast feeders, and Matrigel replated on feeders. At the outset, we profiled and quantified their differentiation efficiency, transcriptome, transcription factor binding sites and DNA-methylation. Subsequent genome-wide analyses allowed us to reconstruct the relevant interactome, thereby forming the regulatory basis for implicating the contrasting differentiation efficiency of the culture conditions. We hypothesized that the parental expressions of FOXC1, FOXD1 and FOXQ1 transcription factors (TFs) are correlative with eventual cardiomyogenic outcome. Through WNT induction of the FOX TFs, we observed the co-activation of WNT3 and EOMES which are potent inducers of mesoderm differentiation. The result strengthened our hypothesis on the regulatory role of the FOX TFs in enhancing mesoderm differentiation capacity of hESCs. Importantly, the final proportions of cells expressing cardiac markers were directly correlated to the strength of FOX inductions within 72 hours after initiation of differentiation across different cell lines and protocols. Thus, we affirmed the relationship between early FOX TF expressions and cardiomyogenesis efficiency.

Mechanistic elements and critical factors of cellular reprogramming revealed by stepwise global gene expression analyses.

A better understanding of the cellular and molecular mechanisms involved in the reprogramming of somatic cells is essential for further improvement of induced pluripotent stem (iPS) cell technology. In this study, we enriched for cells actively undergoing reprogramming at different time points by sorting the cells stained with a stem cell-selective fluorescent chemical probe CDy1 for their global gene expression analysis. Day-to-day comparison of differentially expressed genes showed highly dynamic and transient gene expressions during reprogramming, which were largely distinct from those of fully-reprogrammed cells. An unbiased analysis of functional regulation indicated robust modulation of concurrent programs at critical junctures. Globally, transcriptional programs involved in cell proliferation, morphology and signal transduction were instantly triggered as early as 3 days-post-infection to prepare the cell for reprogramming but became somewhat muted in the final iPS cells. On the other hand, the highly coordinated metabolic reprogramming process was more gradually activated. Subsequent network analysis of differentially expressed genes indicated PDGF-BB as a core player in reprogramming which was verified by our gain- and loss-of-function experiments. As such, our study has revealed previously-unknown insights into the mechanisms of cellular reprogramming.

Patterns of cis-element enrichment reveal potential regulatory modules involved in the transcriptional regulation of anoxia response of japonica rice.

Unlike other cereal species, rice is able to germinate and elongate under anoxia. The regulatory mechanism that configures the transcriptome of rice during anaerobic germination is yet to be established. In this study, the major regulatory modules among anoxia-responsive genes in rice identified from published microarray data were predicted by ab initio analysis of cis-regulatory information content. Statistically overrepresented sequence motifs were detected from bona fide promoter sequences [-1000 to +200], revealing various patterns of cis-element enrichment that are highly correlated with bZIP, ERF and MYB types of transcription factors. As implied by the cis-element enrichment patterns, combinatorial mechanisms configure the overall changes in gene expression during anoxic germination and coleoptile elongation. High enrichment of cis-elements associated with ARF, bZIP, ERF, MYB and WRKY (SUSIBA2) transcription factors was also detected among the glycolytic and fermentative associated genes that were upregulated during anoxia. The patterns established from the global analysis of cis-element distribution for upregulated and downregulated genes and their associations with potential cognate regulatory transcription factors indicate the significant roles of ethylene and abscisic acid mediated signaling during coleoptile elongation under anoxia. In addition, the regulation of genes encoding enzymes in the glycolytic and fermentative metabolism could be associated with abscisic acid and auxin in rice coleoptiles to maintain sugar and ATP levels for longer survival.

Enhanced production of neuroprogenitors, dopaminergic neurons, and identification of target genes by overexpression of sonic hedgehog in human embryonic stem cells.

Molecular and cellular signaling pathways are involved in the process of neural differentiation from human embryonic stem cells (hESC) to terminally differentiated neurons. The Sonic hedgehog (SHH) morphogen is required to direct the differentiation of hESC to several neural subtypes, for example, dopaminergic (DA) or motor neurons. However, the roles of SHH signaling and the pathway target genes that regulate the diversity of cellular responses arising from SHH activation during neurogenesis of hESC have yet to be elucidated. In this study, we report that overexpression of SHH in hESC promotes the derivation of neuroprogenitors (NP), increases proliferation of NP, and subsequently increases the yield of DA neurons. Next, gene expression changes resulting from the overexpression of SHH in hESC-derived NP were examined by genome-wide transcriptional profiling. Categorizing the differentially expressed genes according to the Gene Ontology biological processes showed that they are involved in numerous cellular processes, including neural development, NP proliferation, and neural specification. In silico GLI-binding sites analysis of the differentially expressed genes also identified a set of putative novel direct target genes of SHH in hESC-derived NP, which are involved in nervous system development. Electrophoretic mobility shift assays and promoter-luciferase assays confirmed that GLI1 binds to the promoter region and activates transcription of HEY2, a NOTCH signaling target gene. Taken together, our data provide evidence for the first time that there is cross-talk between the NOTCH and SHH signaling pathways in hESC-derived NP and also provide significant new insights into transcriptional targets in SHH-mediated neural differentiation of hESC.

Integrated transcriptome and binding sites analysis implicates E2F in the regulation of self-renewal in human pluripotent stem cells.

Rapid cellular growth and multiplication, limited replicative senescence, calibrated sensitivity to apoptosis, and a capacity to differentiate into almost any cell type are major properties that underline the self-renewal capabilities of human pluripotent stem cells (hPSCs). We developed an integrated bioinformatics pipeline to understand the gene regulation and functions involved in maintaining such self-renewal properties of hPSCs compared to matched fibroblasts. An initial genome-wide screening of transcription factor activity using in silico binding-site and gene expression microarray data newly identified E2F as one of major candidate factors, revealing their significant regulation of the transcriptome. This is underscored by an elevated level of its transcription factor activity and expression in all tested pluripotent stem cell lines. Subsequent analysis of functional gene groups demonstrated the importance of the TFs to self-renewal in the pluripotency-coupled context; E2F directly targets the global signaling (e.g. self-renewal associated WNT and FGF pathways) and metabolic network (e.g. energy generation pathways, molecular transports and fatty acid metabolism) to promote its canonical functions that are driving the self-renewal of hPSCs. In addition, we proposed a core self-renewal module of regulatory interplay between E2F and, WNT and FGF pathways in these cells. Thus, we conclude that E2F plays a significant role in influencing the self-renewal capabilities of hPSCs.

Inferring transcription factor targets from gene expression changes and predicted promoter occupancy.

We have developed a method for inferring condition-specific targets of transcription factors based on ranking genes by gene expression change and ranking genes based on predicted transcription factor occupancy. The average of these two ranks, used as a test statistic, allows target genes to be inferred in a stringent manner. The method complements chromatin immunoprecipitation experiments by predicting targets under many conditions for which ChIP experiments have not been performed. We used the method to predict targets of 102 yeast transcription factors in approximately 1600 expression microarray experiments. The reliability of the method is suggested by the strong enrichment of genes previously shown to be bound, by the validation of binding to novel targets, by the way transcription factors with similar specificities can be functionally distinguished, and by the greater-than-expected number of regulatory network motifs, such as auto-regulatory interactions, that arise from new, predicted interactions. The combination of ChIP data and the targets inferred from this analysis results in a high-confidence regulatory network that includes many novel interactions. Interestingly, we find only a weak association between conditions in which we can infer the activity of a transcription factor and conditions in which the transcription gene itself is regulated. Thus, methods that rely on transcription factor regulation to help define regulatory interactions may miss regulatory relationships that are detected by the method reported here.

Inferring direct regulatory targets of a transcription factor in the DREAM2 challenge.

In the DREAM2 community-wide experiment on regulatory network inference, one of the challenges was to identify which genes, in a list of 200, are direct regulatory targets of the transcription factor BCL6. The organizers of the challenge defined targets based on gene expression and chromatin immunoprecipitation experiments (ChIP-chip). The expression data were publicly available; the ChIP-chip data were not. In order to assess the likelihood that a gene is a BCL6 target, we used three classes of information: expression-level differences, over-representation of sequence motifs in promoter regions, and gene ontology annotations. A weight was attached to each analysis based on how well it identified BCL6-bound genes as defined by publicly available ChIP-chip data. By the organizers' criteria, our group, GenomeSingapore, performed best. However, our retrospective analysis indicates that this success was dominated by a gene expression analysis that was predicated on a regulatory model known to be favored by the organizers. We also noted that the 200-gene test set was enriched only in genes that are upregulated, while genes bound by BCL6 are enriched in both upregulated and downregulated genes. Together, these observations suggest possible model biases in the selection of the gold-standard gene set and imply that our success was attained in part by adhering to the same assumptions. We argue that model biases of this type are unavoidable in the inference of regulatory networks and, for that reason, we suggest that future community-wide experiments of this type should focus on the prediction of data, rather than models.