Educating the Next Generation of Undergraduate URM Cancer Scientists: Results and Lessons Learned from a Cancer Research Partnership Scholar Program.

To improve cancer disparities among under-represented minority (URM) populations, better representation of URM individuals in cancer research is needed. The San Diego State University and University of California San Diego Moores Cancer Center Partnership is addressing cancer disparities through an educational program targeting undergraduate URM students. The Partnership provides a paid intensive summer research internship enriched with year-round activities that include educational sessions, a journal club, mentorship, social activities, and poster sessions and presentations. Program evaluation through follow-up surveys, focus groups, and other formal and informal feedback, including advisory and program steering committees, are used to improve the program. Long-term follow-up among scholars (minimum of 10 years) provides data to evaluate the program's long-term impact on scholars' education and career path. Since 2016, 63 URM undergraduate students participated in the scholar program. At the year-2 follow-up (2016 cohort; n = 12), 50% had completed their Graduate Record Examination (GRE) and/or applied to graduate or medical school. Lessons learned during the course of the program led to implementation of changes to provide a better learning experience and increase overall program satisfaction. Updates were made to recruitment timeline, improvements of the recruitment processes, refinement of the program contracts and onboarding meetings, identification of essential program coordinator skills and responsibilities, adjustments to program components, and establishment of a well-mapped and scheduled evaluation plan. The Partnership identified best practices and lessons learned for implementing lab-based internship scholar programs in biomedical and public health fields that could be considered in other programs.

Genetic barcoding with fluorescent proteins for multiplexed applications.

Fluorescent proteins, fluorescent dyes and fluorophores in general have revolutionized the field of molecular cell biology. In particular, the discovery of fluorescent proteins and their genes have enabled the engineering of protein fusions for localization, the analysis of transcriptional activation and translation of proteins of interest, or the general tracking of individual cells and cell populations. The use of fluorescent protein genes in combination with retroviral technology has further allowed the expression of these proteins in mammalian cells in a stable and reliable manner. Shown here is how one can utilize these genes to give cells within a population of cells their own biosignature. As the biosignature is achieved with retroviral technology, cells are barcoded 'indefinitely'. As such, they can be individually tracked within a mixture of barcoded cells and utilized in more complex biological applications. The tracking of distinct populations in a mixture of cells is ideal for multiplexed applications such as discovery of drugs against a multitude of targets or the activation profile of different promoters. The protocol describes how to elegantly develop and amplify barcoded mammalian cells with distinct genetic fluorescent markers, and how to use several markers at once or one marker at different intensities. Finally, the protocol describes how the cells can be further utilized in combination with cell-based assays to increase the power of analysis through multiplexing.

Bacterial induction of Snail1 contributes to blood-brain barrier disruption.

Bacterial meningitis is a serious infection of the CNS that results when blood-borne bacteria are able to cross the blood-brain barrier (BBB). Group B Streptococcus (GBS) is the leading cause of neonatal meningitis; however, the molecular mechanisms that regulate bacterial BBB disruption and penetration are not well understood. Here, we found that infection of human brain microvascular endothelial cells (hBMECs) with GBS and other meningeal pathogens results in the induction of host transcriptional repressor Snail1, which impedes expression of tight junction genes. Moreover, GBS infection also induced Snail1 expression in murine and zebrafish models. Tight junction components ZO-1, claudin 5, and occludin were decreased at both the transcript and protein levels in hBMECs following GBS infection, and this repression was dependent on Snail1 induction. Bacteria-independent Snail1 expression was sufficient to facilitate tight junction disruption, promoting BBB permeability to allow bacterial passage. GBS induction of Snail1 expression was dependent on the ERK1/2/MAPK signaling cascade and bacterial cell wall components. Finally, overexpression of a dominant-negative Snail1 homolog in zebrafish elevated transcription of tight junction protein-encoding genes and increased zebrafish survival in response to GBS challenge. Taken together, our data support a Snail1-dependent mechanism of BBB disruption and penetration by meningeal pathogens.

A Multiplexed Cell-Based Assay for the Identification of Modulators of Pre-Membrane Processing as a Target against Dengue Virus.

The DenV pre-membrane protein (prM) is a crucial chaperone for the viral envelope protein, preventing premature fusion with vesicles during viral export. prM molecules in immature particles are cleaved by host proteases, leading to mature fusogenic virions. Blockade of prM cleavage would restrict fusion and represents a novel druggable opportunity against DenV. We have thus established a cell-based platform to monitor prM processing that relies on an engineered two-tag scaffold that travels to the cell surface through the secretory pathway. The assay discriminates between a single cell-surface tag when prM is cleaved and two tags when it is not, as detected through fluorescent-coupled antibodies by flow cytometry. The assay, miniaturized into a 96-well plate format, was multiplexed with the HIV-1 envelope boundary, also cleaved in the same pathway. A pilot screen against 1280 compounds was executed, leading to the identification of a potential active and corroborating the robustness of our assay for large-scale screening. We describe for the first time a cell-based assay that monitors DenV prM processing within the classical secretory pathway, which was exploited to identify a potential novel drug against DenV.

A Single-Cell Platform for Monitoring Viral Proteolytic Cleavage in Different Cellular Compartments.

Infectious diseases affect human health despite advances in biomedical research and drug discovery. Among these, viruses are especially difficult to tackle due to the sudden transfer from animals to humans, high mutational rates, resistance to current treatments, and the intricacies of their molecular interactions with the host. As an example of these interactions, we describe a cell-based approach to monitor specific proteolytic events executed by either the viral-encoded protease or by host proteins on the virus. We then emphasize the significance of examining proteolysis within the subcellular compartment where cleavage occurs naturally. We show the power of stable expression, highlighting the usefulness of the cell-based multiplexed approach, which we have adapted to two independent assays previously developed to monitor (a) the activity of the HIV-1-encoded protease or (b) the cleavage of the HIV-1-encoded envelope protein by the host. Multiplexing was achieved by mixing cells each carrying a different assay or, alternatively, by engineering cells expressing two assays. Multiplexing relies on the robustness of the individual assays and their clear discrimination, further enhancing screening capabilities in an attempt to block proteolytic events required for viral infectivity and spread.

Evolution of TNF-induced apoptosis reveals 550 My of functional conservation.

The Precambrian explosion led to the rapid appearance of most major animal phyla alive today. It has been argued that the complexity of life has steadily increased since that event. Here we challenge this hypothesis through the characterization of apoptosis in reef-building corals, representatives of some of the earliest animals. Bioinformatic analysis reveals that all of the major components of the death receptor pathway are present in coral with high-predicted structural conservation with Homo sapiens. The TNF receptor-ligand superfamilies (TNFRSF/TNFSF) are central mediators of the death receptor pathway, and the predicted proteome of Acropora digitifera contains more putative coral TNFRSF members than any organism described thus far, including humans. This high abundance of TNFRSF members, as well as the predicted structural conservation of other death receptor signaling proteins, led us to wonder what would happen if corals were exposed to a member of the human TNFSF (HuTNFα). HuTNFα was found to bind directly to coral cells, increase caspase activity, cause apoptotic blebbing and cell death, and finally induce coral bleaching. Next, immortalized human T cells (Jurkats) expressing a functional death receptor pathway (WT) and a corresponding Fas-associated death domain protein (FADD) KO cell line were exposed to a coral TNFSF member (AdTNF1) identified and purified here. AdTNF1 treatment resulted in significantly higher cell death (P < 0.0001) in WT Jurkats compared with the corresponding FADD KO, demonstrating that coral AdTNF1 activates the H. sapiens death receptor pathway. Taken together, these data show remarkable conservation of the TNF-induced apoptotic response representing 550 My of functional conservation.

Fluorescent genetic barcoding in mammalian cells for enhanced multiplexing capabilities in flow cytometry.

The discovery of the green fluorescent protein from Aequorea victoria has revolutionized the field of cell and molecular biology. Since its discovery a growing panel of fluorescent proteins, fluorophores and fluorescent-coupled staining methodologies, have expanded the analytical capabilities of flow cytometry. Here, we exploit the power of genetic engineering to barcode individual cells with genes encoding fluorescent proteins. For genetic engineering, we utilize retroviral technology, which allows for the expression of ectopic genetic information in a stable manner in mammalian cells. We have genetically barcoded both adherent and nonadherent cells with different fluorescent proteins. Multiplexing power was increased by combining both the number of distinct fluorescent proteins, and the fluorescence intensity in each channel. Moreover, retroviral expression has proven to be stable for at least a 6-month period, which is critical for applications such as biological screens. We have shown the applicability of fluorescent barcoded multiplexing to cell-based assays that rely themselves on genetic barcoding, or on classical staining protocols. Fluorescent genetic barcoding gives the cell an inherited characteristic that distinguishes it from its counterpart. Once cell lines are developed, no further manipulation or staining is required, decreasing time, nonspecific background associated with staining protocols, and cost. The increasing number of discovered and/or engineered fluorescent proteins with unique absorbance/emission spectra, combined with the growing number of detection devices and lasers, increases multiplexing versatility, making fluorescent genetic barcoding a powerful tool for flow cytometry-based analysis.

Coxsackievirus B exits the host cell in shed microvesicles displaying autophagosomal markers.

Coxsackievirus B3 (CVB3), a member of the picornavirus family and enterovirus genus, causes viral myocarditis, aseptic meningitis, and pancreatitis in humans. We genetically engineered a unique molecular marker, "fluorescent timer" protein, within our infectious CVB3 clone and isolated a high-titer recombinant viral stock (Timer-CVB3) following transfection in HeLa cells. "Fluorescent timer" protein undergoes slow conversion of fluorescence from green to red over time, and Timer-CVB3 can be utilized to track virus infection and dissemination in real time. Upon infection with Timer-CVB3, HeLa cells, neural progenitor and stem cells (NPSCs), and C2C12 myoblast cells slowly changed fluorescence from green to red over 72 hours as determined by fluorescence microscopy or flow cytometric analysis. The conversion of "fluorescent timer" protein in HeLa cells infected with Timer-CVB3 could be interrupted by fixation, suggesting that the fluorophore was stabilized by formaldehyde cross-linking reactions. Induction of a type I interferon response or ribavirin treatment reduced the progression of cell-to-cell virus spread in HeLa cells or NPSCs infected with Timer-CVB3. Time lapse photography of partially differentiated NPSCs infected with Timer-CVB3 revealed substantial intracellular membrane remodeling and the assembly of discrete virus replication organelles which changed fluorescence color in an asynchronous fashion within the cell. "Fluorescent timer" protein colocalized closely with viral 3A protein within virus replication organelles. Intriguingly, infection of partially differentiated NPSCs or C2C12 myoblast cells induced the release of abundant extracellular microvesicles (EMVs) containing matured "fluorescent timer" protein and infectious virus representing a novel route of virus dissemination. CVB3 virions were readily observed within purified EMVs by transmission electron microscopy, and infectious virus was identified within low-density isopycnic iodixanol gradient fractions consistent with membrane association. The preferential detection of the lipidated form of LC3 protein (LC3 II) in released EMVs harboring infectious virus suggests that the autophagy pathway plays a crucial role in microvesicle shedding and virus release, similar to a process previously described as autophagosome-mediated exit without lysis (AWOL) observed during poliovirus replication. Through the use of this novel recombinant virus which provides more dynamic information from static fluorescent images, we hope to gain a better understanding of CVB3 tropism, intracellular membrane reorganization, and virus-associated microvesicle dissemination within the host.

MitoTimer: a novel tool for monitoring mitochondrial turnover.

Fluorescent Timer, or DsRed1-E5, is a mutant of the red fluorescent protein, dsRed, in which fluorescence shifts over time from green to red as the protein matures. This molecular clock gives temporal and spatial information on protein turnover. To visualize mitochondrial turnover, we targeted Timer to the mitochondrial matrix with a mitochondrial-targeting sequence (coined "MitoTimer") and cloned it into a tetracycline-inducible promoter construct to regulate its expression. Here we report characterization of this novel fluorescent reporter for mitochondrial dynamics. Tet-On HEK 293 cells were transfected with pTRE-tight-MitoTimer and production was induced with doxycycline (Dox). Mitochondrial distribution was demonstrated by fluorescence microscopy and verified by subcellular fractionation and western blot analysis. Dox addition for as little as 1 h was sufficient to induce MitoTimer expression within 4 h, with persistence in the mitochondrial fraction for up to 6 d. The color-specific conformation of MitoTimer was stable after fixation with 4% paraformaldehyde. Ratiometric analysis of MitoTimer revealed a time-dependent transition from green to red over 48 h and was amenable to analysis by fluorescence microscopy and flow cytometry of whole cells or isolated mitochondria. A second Dox administration 48 h after the initial induction resulted in a second round of expression of green MitoTimer. The extent of new protein incorporation during a second pulse was increased by administration of a mitochondrial uncoupler or simvastatin, both of which trigger mitophagy and biogenesis. MitoTimer is a novel fluorescent reporter protein that can reveal new insights into mitochondrial dynamics within cells. Coupled with organelle flow cytometry, it offers new opportunities to investigate mitochondrial subpopulations by biochemical or proteomic methods.

Modular construction of quaternary hemiaminal-based inhibitor candidates and their in cellulo assessment with HIV-1 protease.

Non-peptidomimetic drug-like protease inhibitors have potential for circumventing drug resistance. We developed a much-improved synthetic route to our previously reported inhibitor candidate displaying an unusual quaternized hemi-aminal. This functional group forms from a linear precursor upon passage into physiological media. Seven variants were prepared and tested in cellulo with our HIV-1 fusion-protein technology that result in an eGFP-based fluorescent readout. Three candidates showed inhibition potency above 20µM and toxicity at higher concentrations, making them attractive targets for further refinement. Importantly, our class of original inhibitor candidates is not recognized by two major multidrug resistance pumps, quite in contrast to most clinically applied HIV-1 protease inhibitors.

A Novel Two-Tag System for Monitoring Transport and Cleavage through the Classical Secretory Pathway - Adaptation to HIV Envelope Processing.

The classical secretory pathway is essential for the transport of a host of proteins to the cell surface and/or extracellular matrix. While the pathway is well-established, many factors still remain to be elucidated. One of the most relevant biological processes that occur during transport involves the cleavage of pro-proteins by enzymes residing in the endoplasmic reticulum/Golgi/TransGolgi Network compartment. Teasing out the requirements involved in the classical secretory pathway and cleavage during transport would shed new light into mis-regulation leading to disease. Current methodologies fail to link transport and cleavage at the single cell level. Here, we describe a cell-based assay that relies on an engineered protein scaffold that can discriminate between transport to the cell surface, in the absence or presence of cleavage. Our novel two-tag system works in a robust and quantitative manner and distinguishes between cleaved and non-cleaved events based on cell surface expression of one or two epitope tags, respectively. Here, we have used the HIV-1 envelope as a substrate, which is cleaved during transport, as proof of principle. Importantly, this assay can be easily coupled to existing siRNA-based screens to identify novel regulators and effectors involved in transport and/or cleavage of cell surface proteins. In addition, unlike other in vivo based assays, the assay described here can also be easily adapted to drug discovery purposes.

Bacteriophage adhering to mucus provide a non-host-derived immunity.

Mucosal surfaces are a main entry point for pathogens and the principal sites of defense against infection. Both bacteria and phage are associated with this mucus. Here we show that phage-to-bacteria ratios were increased, relative to the adjacent environment, on all mucosal surfaces sampled, ranging from cnidarians to humans. In vitro studies of tissue culture cells with and without surface mucus demonstrated that this increase in phage abundance is mucus dependent and protects the underlying epithelium from bacterial infection. Enrichment of phage in mucus occurs via binding interactions between mucin glycoproteins and Ig-like protein domains exposed on phage capsids. In particular, phage Ig-like domains bind variable glycan residues that coat the mucin glycoprotein component of mucus. Metagenomic analysis found these Ig-like proteins present in the phages sampled from many environments, particularly from locations adjacent to mucosal surfaces. Based on these observations, we present the bacteriophage adherence to mucus model that provides a ubiquitous, but non-host-derived, immunity applicable to mucosal surfaces. The model suggests that metazoan mucosal surfaces and phage coevolve to maintain phage adherence. This benefits the metazoan host by limiting mucosal bacteria, and benefits the phage through more frequent interactions with bacterial hosts. The relationships shown here suggest a symbiotic relationship between phage and metazoan hosts that provides a previously unrecognized antimicrobial defense that actively protects mucosal surfaces.

Protocol for a mammalian cell-based assay for monitoring the HIV-1 protease activity.

Proteases are essential at different stages of the viral life cycle and for the establishment of a successful infection. Monitoring the catalytic activity of proteases in an easy and straightforward manner can thus drastically facilitate the discovery of novel antivirals, as well as help elucidate the activity and mechanism of action of the viral protease under study. In our laboratory, we have developed an assay in T-cells with a robust read-out to monitor the proteolytic activity of HIV-1 Protease (PR). The assay utilizes the prototypic transcription factor Gal4, which consists of the N-terminal DNA-binding domain and the C-terminal trans-activation domain. The assay is based upon (1) introduction of PR in between the two Gal4 domains to obtain a PR/Gal4 fusion protein and (2) utilization of the enhanced Green Fluorescent Protein as reporter of PR activity.In order to overcome the possible cellular cytotoxicity of PR, the fusion protein in our assay is under the control of a tetracycline-inducible promoter. This ensures that it will be expressed only when needed, upon the addition of tetracycline or doxycycline. When active, PR has autocatalytic activity and cleaves itself from the Gal4 domains, resulting in the inability to induce eGFP expression. However, if PR activity is blocked or it is inactive, the two domains remain intact, resulting in eGFP expression. The assay can therefore be utilized to analyze the inhibitory effects of factors, peptides or compounds, designed on a rational- or nonrational-based approach, in the natural milieu of infection, where eGFP serves as a biosensor for PR activity.

Purification of the COP9 signalosome complex and binding partners from human T cells.

The COP9 Signalosome (CSN) is a highly conserved eight subunit protein complex associated with a wide range of essential biological functions in eukaryotic cells, and directly involved in processes including deneddylation, phosphorylation, and ubiquitination. Despite its significant role, very few studies have been undertaken to reveal the interactions between the CSN and its binding partners, and none in human T cells. Here we present a purification method for the CSN and binding proteins via the Streptavidin-Binding Peptide (SBP) fused to CSN Subunit 1 (CSN1). Using this method, coupled with liquid chromatography-mass spectrometry analysis, we identified all eight subunits of the CSN, as well as expected and putative novel binding partners such as a tumor suppressor under the control of Cullin4a-ligase complex; Neurofibromin 2 (Merlin). This work presents a method for fast, reliable, and specific affinity-based purification of a protein complex from a nonadherent cell line. The purification of the CSN and binding partners from T cells can elucidate the roles of CSN in a cell type where it has never been studied before. This proteomic-based approach can broaden our understanding of the functions of the CSN in contexts such as viral-host interactions or immune activation in their natural milieu.

An assay to monitor HIV-1 protease activity for the identification of novel inhibitors in T-cells.

The emergence of resistant HIV strains, together with the severe side-effects of existing drugs and lack of development of effective anti-HIV vaccines highlight the need for novel antivirals, as well as innovative methods to facilitate their discovery. Here, we have developed an assay in T-cells to monitor the proteolytic activity of the HIV-1 protease (PR). The assay is based on the inducible expression of HIV-1 PR fused within the Gal4 DNA-binding and transactivation domains. The fusion protein binds to the Gal4 responsive element and activates the downstream reporter, enhanced green fluorescent protein (eGFP) gene only in the presence of an effective PR Inhibitor (PI). Thus, in this assay, eGFP acts as a biosensor of PR activity, making it ideal for flow cytometry based screening. Furthermore, the assay was developed using retroviral technology in T-cells, thus providing an ideal environment for the screening of potential novel PIs in a cell-type that represents the natural milieu of HIV infection. Clones with the highest sensitivity, and robust, reliable and reproducible reporter activity, were selected. The assay is easily adaptable to other PR variants, a multiplex platform, as well as to high-throughput plate reader based assays and will greatly facilitate the search for novel peptide and chemical compound based PIs in T-cells.

Roles for endoplasmic reticulum-associated degradation and the novel endoplasmic reticulum stress response gene Derlin-3 in the ischemic heart.

RATIONALE: Stresses, such as ischemia, impair folding of nascent proteins in the rough endoplasmic reticulum (ER), activating the unfolded protein response, which restores efficient ER protein folding, thus leading to protection from stress. In part, the unfolded protein response alleviates ER stress and cell death by increasing the degradation of terminally misfolded ER proteins via ER-associated degradation (ERAD). ERAD is increased by the ER stress modulator, activating transcription factor (ATF)6, which can induce genes that encode components of the ERAD machinery. OBJECTIVE: Recently, it was shown that the mouse heart is protected from ischemic damage by ATF6; however, ERAD has not been studied in the cardiac context. A recent microarray study showed that the Derlin-3 (Derl3) gene, which encodes an important component of the ERAD machinery, is robustly induced by ATF6 in the mouse heart. METHODS AND RESULTS: In the present study, activated ATF6 induced Derl3 in cultured cardiomyocytes, and in the heart, in vivo. Simulated ischemia (sI), which activates ER stress, induced Derl3 in cultured myocytes, and in an in vivo mouse model of myocardial infarction, Derl3 was also induced. Derl3 overexpression enhanced ERAD and protected cardiomyocytes from simulated ischemia-induced cell death, whereas dominant-negative Derl3 decreased ERAD and increased simulated ischemia-induced cardiomyocyte death. CONCLUSIONS: This study describes a potentially protective role for Derl3 in the heart, and is the first to investigate the functional consequences of enhancing ERAD in the cardiac context.

NF-kappaB signaling, elastase localization, and phagocytosis differ in HIV-1 permissive and nonpermissive U937 clones.

To identify positive or negative factors for HIV-1 infectivity, clones from the U937 promonocytic cell line that express similar levels of CD4 and CXCR4, but differ in HIV-1 susceptibility, were compared. In contrast to HIV-1 permissive clone 10 (plus), nonpermissive clone 17 (minus) was adherent to coverslips coated with chemokines, was phagocytic, killed bacteria, and expressed human leukocyte elastase (HLE) in a granule-like compartment (HLEG) that was never detected at the cell surface (HLECS). In contrast to the minus clone, the plus clone expressed HLE on the cell surface and was adherent to coverslips coated with the HLECS ligands alpha1proteinase inhibitor (alpha1PI, alpha1antitrypsin) and the HIV-1 fusion peptide. The phosphorylation status of several important signaling proteins was studied at the single cell level. Tumor suppressor p53, NF-kappaB p65, and Akt were constitutively phosphorylated in the plus clone, but not in the minus clone. Surprisingly, both alpha1PI and LPS induced phosphorylation of NF-kappaB p65 Ser-536 in both clones, but induced dephosphorylation of Ser-529 in the plus clone only. HIV-1 permissivity was conferred to the minus clone in a manner that required stimulation by both alpha1PI and LPS and was coincident to NF-kappaB p65 phosphorylation/dephosphorylation events as well as translocation of HLE to the cell surface. Even when stimulated, the minus clone exhibited greater reverse transcriptase activity, but less p24, than the plus clone. Results presented suggest that HIV-1 uptake and production efficiency are influenced by signaling profiles, receptor distribution, and the phagocytic capacity specific to the stage of differentiation of the CD4+ target cell.

Accurate prediction of HIV-1 drug response from the reverse transcriptase and protease amino acid sequences using sparse models created by convex optimization.

MOTIVATION: Genotype-phenotype modeling problems are often overcomplete, or ill-posed, since the number of potential predictors-genes, proteins, mutations and their interactions-is large relative to the number of measured outcomes. Such datasets can still be used to train sparse parameter models that generalize accurately, by exerting a principle similar to Occam's Razor: When many possible theories can explain the observations, the most simple is most likely to be correct. We apply this philosophy to modeling the drug response of Type-1 Human Immunodeficiency Virus (HIV-1). Owing to the decreasing expense of genetic sequencing relative to in vitro phenotype testing, a statistical model that reliably predicts viral drug response from genetic data is an important tool in the selection of antiretroviral therapy (ART). The optimization techniques described will have application to many genotype-phenotype modeling problems for the purpose of enhancing clinical decisions. RESULTS: We describe two regression techniques for predicting viral phenotype in response to ART from genetic sequence data. Both techniques employ convex optimization for the continuous subset selection of a sparse set of model parameters. The first technique, the least absolute shrinkage and selection operator, uses the l(1) norm loss function to create a sparse linear model; the second, the support vector machine with radial basis kernel functions, uses the epsilon-insensitive loss function to create a sparse non-linear model. The techniques are applied to predict the response of the HIV-1 virus to 10 reverse transcriptase inhibitor and 7 protease inhibitor drugs. The genetic data are derived from the HIV coding sequences for the reverse transcriptase and protease enzymes. When tested by cross-validation with actual laboratory measurements, these models predict drug response phenotype more accurately than models previously discussed in the literature, and other canonical techniques described here. Key features of the methods that enable this performance are the tendency to generate simple models where many of the parameters are zero, and the convexity of the cost function, which assures that we can find model parameters to globally minimize the cost function for a particular training dataset. AVAILABILITY: Results, tables and figures are available at ftp://ftp.genesecurity.net. SUPPLEMENTARY INFORMATION: An Appendix to accompany this article is available at Bioinformatics online.