Beaulaurier J, Schadt EE & Fang G#, Deciphering bacterial epigenomes using modern sequencing technologies, Nature Reviews Genetics, 2018 (link)

Beaulaurier J, Zhu S, Deikus G, Mogno I, Zhang XS, Davis-Richardson A, Canepa R, Triplett EW, Faith JJ, Sebra R, Schadt EE & Fang G#,  Metagenomic binning and association of plasmids with bacterial host genomes using DNA methylation, Nature Biotechnology, 10.1038/nbt.4037​, 2018 (link)

Zhu S, Beaulaurier J, Deikus G, Wu T, Strahl M, Hao Z, Luo G, Gregory JA, Chess A, He C, Xiao A, Sebra R, Schadt EE, Fang G#, Mapping and characterizing N6-methyladenine in eukaryotic genomes using single molecule real-time sequencing, Genome Research, doi: 10.1101/gr.231068.117, 2018 (link)

Wu TP, Wang T, Seetin MG, Lai Y, Zhu S, Lin K, Liu Y, Byrum SD, Mackintosh SG, Zhong M, Tackett A, Wang G, Hon LS, Fang G, Swenberg J & Xiao A, DNA methylation on N6-adenine in mammalian embryonic stem cells, Nature, 10.1038/nature17640, 2016 (link)

Beaulaurier J, Zhang XS, Zhu S, Sebra R, Rosenbluh C, Deikus G, Shen N, Munera D, Waldor MK, Blaser MJ, Chess A, Schadt EE# & Fang G#, Single molecule-level detection and long read-based phasing of epigenetic variations in bacterial methylomes, Nature Communications, 10.1038/ncomms8438, 2015. (link)

Fang G*, Munera D*, Friedman DI, Mandlik A, Chao MC, Banerjee O, Feng Z, Losic B, Mahajan MC, Jabado OJ, Deikus G, et al. , Genome-wide map of methylated adenine residues using single-molecule real-time sequencing in pathogenic Escherichia coli, Nature Biotechnology, 10.1038/nbt.2432, 2012. (link)


Pedro H. Oliveira, Alex Kim, Ognjen Sekulovic, Elizabeth M. Garrett, Dominika Trzilova, Edward A. Mead, Theodore Pak, Shijia Zhu, Gintaras Deikus, Marie Touchon, Colleen Beckford, Nathalie E. Zeitouni, Deena Altman, Elizabeth Webster, Irina Oussenko, Aneel K. Aggarwal, Ali Bashir, Gopi Patel, Camille Hamula, Shirish Huprikar, Richard J. Roberts, Eric E. Schadt, Robert Sebra, Harm van Bakel, Andrew Kasarskis, Rita Tamayo, Aimee Shen, Gang Fang#, Epigenomic landscape of the human pathogen Clostridium difficile, bioRxiv 398891 (link)

Full list (since 2012)

Deciphering bacterial epigenomes using modern sequencing technologies
John Beaulaurier, Eric Schadt & Gang Fang#

Prokaryotic DNA contains three types of methylation: N6-methyladenine,
N4-methylcytosine and 5-methylcytosine. The lack of tools to analyse the frequency and distribution of methylated residues in bacterial genomes has prevented a full understanding of their functions. Now , advances in DNA sequencing technology , including single- molecule, real- time sequencing and nanopore- based sequencing, have provided new opportunities for systematic detection of all three forms of methylated DNA at a genome- wide scale and offer unprecedented opportunities for achieving a more complete understanding of bacterial epigenomes. Indeed, as the number of mapped bacterial methylomes approaches 2,000, increasing evidence supports roles for methylation in regulation of gene expression, virulence and pathogen–host interactions.

Nature Reviews Genetics,​, 2018

Metagenomic binning and association of plasmids with bacterial host genomes using DNA methylation
John Beaulaurier, Shijia Zhu, Gintaras Deikus, Ilaria Mogno, Xue-Song Zhang, Austin Davis-Richardson, Ronald Canepa, Eric Triplett, Jeremiah Faith, Robert Sebra, Eric Schadt & Gang Fang#

Shotgun metagenomics methods enable characterization of microbial communities in human microbiome and environmental samples. Assembly of metagenome sequences does not output whole genomes, so computational binning methods have been developed to cluster sequences into genome ‘bins’. These methods exploit sequence composition, species abundance, or chromosome organization but cannot fully distinguish closely related species and strains. We present a binning method that incorporates bacterial DNA methylation signatures, which are detected using single-molecule real-time sequencing. Our method takes advantage of these endogenous epigenetic barcodes to resolve individual reads and assembled contigs into species- and strain-level bins. We validated our method using synthetic and real microbiome sequences. In addition to genome binning, we show that our method links plasmids and other mobile genetic elements to their host species in a real microbiome sample. Incorporation of DNA methylation information into shotgun metagenomics analyses will complement existing methods to enable more accurate sequence binning. 

Nature Biotechnology, 10.1038/nbt.4037​, 2018

Highlighted in Nature Methods (link)

Media coverage: GEN NewsPacBioGenomeWebMD MagazineBioITWorldScience DailyInfection Control TodayPHYS.

Mapping and characterizing N6-methyladenine in eukaryotic genomes using single molecule real-time sequencing
Shijia Zhu, John Beaulaurier, Gintaras Deikus, Tao Wu, Maya Strahl, Ziyang Hao, Guanzheng Luo, James A Gregory, Andrew Chess, Chuan He, Andrew Xiao, Robert Sebra, Eric E Schadt, and Gang Fang#

N6-methyladenine (m6dA) has been discovered as a novel form of DNA methylation prevalent in eukaryotes, however, methods for high resolution mapping of m6dA events are still lacking. Single-molecule real-time (SMRT) sequencing has enabled the detection of m6dA events at single-nucleotide resolution in prokaryotic genomes, but its application to detecting m6dA in eukaryotic genomes has not been rigorously examined. Herein, we identified unique characteristics of eukaryotic m6dA methylomes that fundamentally differ from those of prokaryotes. Based on these differences, we describe the first approach for mapping m6dA events using SMRT sequencing specifically designed for the study of eukaryotic genomes, and provide appropriate strategies for designing experiments and carrying out sequencing in future studies. We apply the novel approach to study two eukaryotic genomes. For green algae, we construct the first complete genome-wide map of m6dA at single nucleotide and single molecule resolution. For human lymphoblastoid cells (hLCLs), it was necessary to integrate SMRT sequencing data with independent sequencing data. The joint analyses suggest putative m6dA events are enriched in the promoters of young full-length LINE-1 elements (L1s), but call for validation by additional methods. These analyses demonstrate a general method for rigorous mapping and characterization of m6dA events in eukaryotic genomes. 

Genome Research, doi: 10.1101/gr.231068.117, 2018    

DNA methylation on N6-adenine in mammalian embryonic stem cells
Tao P. Wu, Tao Wang, Matthew G. Seetin, Yongquan Lai, Shijia Zhu, Kaixuan Lin, Yifei Liu, Stephanie D. Byrum, Samuel G. Mackintosh, Mei Zhong, Alan Tackett, Guilin Wang, Lawrence S. Hon, Gang Fang, James Swenberg & Andrew Xiao

It has been widely accepted that 5-methylcytosine is the only form of DNA methylation in mammalian genomes. Here we identify N6-methyladenine as another form of DNA modification in mouse embryonic stem cells. Alkbh1 encodes a demethylase for N6-methyladenine. An increase of N6-methyladenine levels in Alkbh1-deficient cells leads to transcriptional silencing. N6-methyladenine deposition is inversely correlated with the evolutionary age of LINE-1 transposons; its deposition is strongly enriched at young (<1.5 million years old) but not old (>6 million years old) L1 elements. The deposition of N6-methyladenine correlates with epigenetic silencing of such LINE-1 transposons, together with their neighbouring enhancers and genes, thereby resisting the gene activation signals during embryonic stem cell differentiation. As young full-length LINE-1 transposons are strongly enriched on the X chromosome, genes located on the X chromosome are also silenced. Thus, N6-methyladenine developed a new role in epigenetic silencing in mammalian evolution distinct from its role in gene activation in other organisms. Our results demonstrate that N6-methyladenine constitutes a crucial component of the epigenetic regulation repertoire in mammalian genomes. 

Nature, 10.1038/nature17640, 2016  

Dysregulation of miRNA-9 in a Subset of Schizophrenia Patient-Derived Neural Progenitor Cells
Aaron Topol*, Shijia Zhu*, Brigham J. Hartley, Jane English, Mads E. Hauberg, Ngoc Tran, Chelsea Ann Rittenhouse, Anthony Simone, Douglas M. Ruderfer, Jessica Johnson, Ben Readhead, Yoav Hadas, Peter A. Gochman, Ying-Chih Wang, Hardik Shah, Gerard Cagney, Judith Rapoport, Fred H. Gage, Joel T. Dudley, Pamela Sklar, Manuel Mattheisen, David Cotter, Gang Fang# & Kristen J. Brennand#

Converging evidence indicates that microRNAs (miRNAs) may contribute to disease risk for schizophrenia (SZ). We show that microRNA-9 (miR-9) is abundantly expressed in control neural progenitor cells (NPCs) but also significantly downregulated in a subset of SZ NPCs. We observed a strong correlation between miR-9 expression and miR-9 regulatory activity in NPCs as well as between miR-9 levels/activity, neural migration, and diagnosis. Overexpression of miR-9 was sufficient to ameliorate a previously reported neural migration deficit in SZ NPCs, whereas knockdown partially phenocopied aberrant migration in control NPCs. Unexpectedly, proteomic- and RNA sequencing (RNA-seq)-based analysis revealed that these effects were mediated primarily by small changes in expression of indirect miR-9 targets rather than large changes in direct miR-9 targets; these indirect targets are enriched for migration-associated genes. Together, these data indicate that aberrant levels and activity of miR-9 may be one of the many factors that contribute to SZ risk, at least in a subset of patients. 

Cell Reports,, 2016  

(*co-first author; #co-corresponding authors)

Single molecule-level detection and long read-based phasing of epigenetic variations in bacterial methylomes
John Beaulaurier, Xue-Song Zhang, Shijia Zhu, Robert Sebra, Chaggai Rosenbluh, Gintaras Deikus, Nan Shen, Diana Munera, Matthew K Waldor, Martin J Blaser, Andrew Chess, Eric E Schadt#, Gang Fang#

Beyond its role in host defense, bacterial DNA methylation also plays important roles in the regulation of gene expression, virulence and antibiotic resistance. Bacterial cells in a clonal population can generate epigenetic heterogeneity to increase population-level phenotypic plasticity. Single molecule, real-time (SMRT) sequencing enables the detection of N6-methyladenine and N4-methylcytosine, two major types of DNA modifications comprising the bacterial methylome. However, existing SMRT sequencing-based methods for studying bacterial methylomes rely on a population-level consensus that lacks the single-cell resolution required to observe epigenetic heterogeneity. Here, we present SMALR (single-molecule modification analysis of long reads), a novel framework for single molecule-level detection and phasing of DNA methylation. Using seven bacterial strains, we show that SMALR yields significantly improved resolution and reveals distinct types of epigenetic heterogeneity. SMALR is a powerful new tool that enables de novo detection of epigenetic heterogeneity and empowers investigation of its functions in bacterial populations. 

Nature Communications, doi:10.1038/ncomms8438, 2015

Press Release;
Media coverage: GEN, GenomeWeb, PHYS, Infection Control, ScienceDaily among others.

A Cytosine Methytransferase Modulates the Cell Envelope Stress Response in the Cholera Pathogen
Michael C. Chao, Shijia Zhu, Satoshi Kimura, Brigid M. Davis, Eric E. Schadt, Gang Fang,# Matthew K. Waldor#

Methylation of DNA is used by numerous organisms to regulate a wide variety of cellular processes, but specific roles for most DNA methyltransferases have not been defined. We studied one such enzyme in Vibrio cholerae, the cholera pathogen, using genome-wide approaches to compare DNA methylation, gene expression, and the sets of genes required or dispensable for growth in bacterial strains that produced or lacked this enzyme. These studies allowed us to identify numerous cellular processes regulated, either directly or indirectly, by this cytosine methyltransferase. In particular, we found that an absence of enzyme activity was associated with reduced levels of a bacterial stress response; consequently, a stress response pathway that is essential in wild type bacteria is not needed for survival of the mutant lacking the methyltransferase. Similar genome-wide analyses can likely to be used to define the cellular roles of many additional uncharacterized DNA methyltransferases. 

PLoS Genetics, doi:10.1371/journal.pgen.1005666, 2015

(#co-corresponding authors)

Genome-wide map of methylated adenine residues using single-molecule real-time sequencing in pathogenic Escherichia coli
Gang Fang, Diana Munera, David I. Friedman, Anjali Mandlik, Michael C. Chao, Onureena Banerjee, Zhixing Feng, Bojan Losic, Milind C. Mahajan, Omar J. Jabado, Gintaras Deikus, Tyson A. Clark, Khai Luong, Iain A. Murray, Brigid M. Davis, Alona Keren-Paz, Andrew Chess, Richard J. Roberts, Jonas Korlach, Steve W. Turner, Vipin Kumar, Matthew K. Waldor, Eric E. Schadt

Single-molecule real-time (SMRT) DNA sequencing allows the systematic detection of chemical modifications such as methylation but has not previously been applied on a genome-wide scale. We used this approach to detect 49,311 putative 6-methyladenine (m6A) residues and 1,407 putative 5-methylcytosine (m5C) residues in the genome of a pathogenic Escherichia coli strain. We obtained strand-specific information for methylation sites and a quantitative assessment of the frequency of methylation at each modified position. We deduced the sequence motifs recognized by the methyltransferase enzymes present in this strain without prior knowledge of their specificity. Furthermore, we found that deletion of a phage-encoded methyltransferase-endonuclease (restriction-modification; RM) system induced global transcriptional changes and led to gene amplification, suggesting that the role of RM systems extends beyond protecting host genomes from foreign DNA. 

Nature Biotechnology, doi:10.1038/nbt.2432, 2012

Also highlighted in Nature Reviews Genetics and
Nature Reviews Microbiology; Media coverage includes: Bio-IT World, TheScientist, PHYS.

Autotransporters but not pAA are critical for rabbit colonization by Shiga toxin-producing Escherichia coli O104:H4
Diana Munera, Jennifer M. Ritchie, Stavroula K. Hatzios, Rod Bronson, Gang Fang, Eric E. Schadt, Brigid M. Davis & Matthew K. Waldor

The outbreak of diarrhoea and haemolytic uraemic syndrome that occurred in Germany in 2011 was caused by a Shiga toxin-producing enteroaggregative Escherichia coli (EAEC) strain. The strain was classified as EAEC owing to the presence of a plasmid (pAA) that mediates a characteristic pattern of aggregative adherence on cultured cells, the defining feature of EAEC that has classically been associated with virulence. Here we describe an infant rabbit-based model of intestinal colonization and diarrhoea caused by the outbreak strain, which we use to decipher the factors that mediate the pathogen's virulence. Shiga toxin is the key factor required for diarrhoea. Unexpectedly, we observe that pAA is dispensable for intestinal colonization and development of intestinal pathology. Instead, chromosome-encoded autotransporters are critical for robust colonization and diarrhoeal disease in this model. Our findings suggest that conventional wisdom linking aggregative adherence to EAEC intestinal colonization is false for at least a subset of strains.

Nature Communications, doi:10.1038/ncomms4080, 2014

Altered WNT Signaling in Human Induced Pluripotent Stem Cell Neural Progenitor Cells Derived from Four Schizophrenia Patients
Aaron Topol, Shijia Zhu, Ngoc Tran, Anthony Simone, Gang Fang, Kristen J. Brennand

Schizophrenia (SZ) is a devastating psychiatric disorder hypothesized to be a neurodevelopmental condition arising as a consequence of dysregulation of brain development. WNT signaling is important for neural patterning, proliferation and migration, and synapse formation; converging postmortem, rodent, and pharmacologic evidence suggests that WNT signaling may contribute to SZ. We used human induced pluripotent stem cell (hiPSC) derived forebrain patterned neural progenitor cells (NPCs) to investigate canonical WNT activity in a pilot cohort of four patients with SZ. Future studies comprising larger patient cohorts are necessary to determine whether aberrant canonical WNT signaling is a causal molecular factor contributing to aberrant neural patterning and neuronal maturation in SZ or simply a noncell autonomous consequence of increased oxidative stress.

Biological Psychiatry, doi: 10.1016/j.biopsych, 2015

Phenotypic differences in hiPSC NPCs derived from patients with schizophrenia
Kristen Brennand, Jeffrey Savas, Yongsung Kim, Ngoc Tran, Anthony Simone, Kazue Hashimoto-Torii, Kristin Beaumont, Hyung Joon Kim, Aaron Topol, Ian Ladran, Mohammed Abdelrahim, Bridget Matikainen-Ankney, Shih-hui Chao, Milan Mrksich, Pasko Rakic, Gang Fang, Bin Zhang, John Yates III, Fred H. Gage

Consistent with recent reports indicating that neurons differentiated in vitro from human-induced pluripotent stem cells (hiPSCs) are immature relative to those in the human brain, gene expression comparisons of our hiPSC-derived neurons to the Allen BrainSpan Atlas indicate that they most resemble fetal brain tissue. This finding suggests that, rather than modeling the late features of schizophrenia (SZ), hiPSC-based models may be better suited for the study of disease predisposition. We now report that a significant fraction of the gene signature of SZ hiPSC-derived neurons is conserved in SZ hiPSC neural progenitor cells (NPCs). We used two independent discovery-based approaches—microarray gene expression and stable isotope labeling by amino acids in cell culture (SILAC) quantitative proteomic mass spectrometry analyses—to identify cellular phenotypes in SZ hiPSC NPCs from four SZ patients. From our findings that SZ hiPSC NPCs show abnormal gene expression and protein levels related to cytoskeletal remodeling and oxidative stress, we predicted, and subsequently observed, aberrant migration and increased oxidative stress in SZ hiPSC NPCs. These reproducible NPC phenotypes were identified through scalable assays that can be applied to expanded cohorts of SZ patients, making them a potentially valuable tool with which to study the developmental mechanisms contributing to SZ.

Molecular Psychiatry, doi: 10.1038/mp.2014.22, 2014

Modeling Kinetic Rate Variation in Third Generation DNA Sequencing Data to Detect Putative Modifications to DNA Bases
Eric E. Schadt*, Onureena Banerjee*, Gang Fang*, Zhixing Feng, Wing H. Wong, Xuegong Zhang, Andrey Kislyuk, Tyson A. Clark, Khai Luong, Vipin Kumar, Alice Chen-Plotkin, Neal Sondheimer, Jonas Korlach, Andrew Kasarskis.

While significant inroads have been made identifying small nucleotide variation and structural variations in DNA that impact phenotypes of interest, progress has not been as dramatic regarding epigenetic changes and base-level damage to DNA, largely due to technological limitations in assaying all known and unknown types of modifications at genome scale. Recently single molecule real time (SMRT) sequencing has been reported to identify kinetic variation (KV) events that have been demonstrated to reflect epigenetic changes of every known type, providing a path forward for detecting base modifications as a routine part of sequencing. However, to date, no statistical framework has been proposed to enhance the power to detect these events while also controlling for false positive events. By modeling enzyme kinetics in the neighborhood of an arbitrary location in a genomic region of interest as a conditional random field, we provide a statistical framework for incorporating kinetic information at a test positions of interest as well as at neighboring sites that help enhance the power to detect KV events. The performance of this and related models is explored, with the best performing model applied to plasmid DNA isolated from Escherichia coli and mitochondrial DNA isolated from human brain tissue. We highlight widespread kinetic variation events, some of which strongly associate with known modification events while others represent putative chemically modified sites of unknown types.   

Genome Research, doi:10.1101/gr.136739.111, 2012 (*co-first authors)

Comprehensive methylome characterization of Mycoplasma genitalium and Mycoplasma pneumoniae, at single-base resolution
Maria Lluch Senar, Khai Luong, Veroica Llorens, Javi Delgado, Gang Fang, Kristi Spittle, Tyson Clark, Eric Schadt, Steve Turner, Jonas Korlach, Luis Serrano

We define the methylome of two closely related bacteria, M. genitalium and M. pneumoniae, by single-molecule real-time (SMRT) DNA sequencing. In M. pneumoniae we found two previously unknown N6-methyladenine methyltransferase specificities, one of which is also found in M. genitalium. The common methyltransferase is a Dam-like methylase, and was attributed to its corresponding gene using cloned plasmids in a methyltransferase-free E. coli strain, while the second methylase is of type I and uniquely present in M. pneumoniae. Analysis of the distribution of methylation sites across the genome of M. pneumoniae at exponential and stationary growth suggests a potential role for methylation in regulating the cell cycle as well as in gene regulation.
PLoS Genetics, 9(1): e1003191. doi:10.1371/journal.pgen.1003191, 2013

Detecting DNA modifications from SMRT sequencing data by modeling sequence context dependence of polymerase kinetic
Zhixing Feng, Gang Fang, Jonas Korlach, Tyson Clark, Khai Luong, Xuegong Zhang, Wing Wong, and Eric Schadt

DNA modications such as methylation and DNA damage can play critical regulatory roles in biological systems. Single molecule, real time (SMRT) sequencing technology generates DNA sequences as well as DNA polymerase kinetic information that can be used for the direct detection of DNA modications. We demonstrate that local sequence context has a strong impact on DNA polymerase kinetics in the neighborhood of the incorporation site during the DNA synthesis reaction, allowing for the possibility of estimating the expected kinetic rate of the enzyme at the incorporation site using kinetic rate information collected from existing SMRT sequencing data (historical data) covering the same local sequence contexts of interest. We develop a Empirical Bayesian hierarchical model for incorporating historical data. Our results show that the model could greatly increase DNA modication detection accuracy, and reduce requirement of control data coverage. For some DNA modications that have a strong signal, a control sample is even not needed by using historical data as alternative to control. Thus, sequencing cost can be greatly reduced by using the model.
PLoS Computational Biology,9(3): e1002935. doi:10.1371/journal.pcbi.1002935, 2013

High-order SNP Combinations Associated with Complex Diseases: Efficient Discovery, Statistical Power and Functional Interactions
Gang Fang*, Majda Haznadar, Wen Wang, Haoyu Yu, Michael Steinbach, Tim Church, William Oetting, Brian Van Ness and Vipin Kumar*.

There has been increased interest in discovering combinations of single-nucleotide polymorphisms (SNPs) that are strongly associated with a phenotype even if each SNP has little individual effect. Efficient approaches have been proposed for searching two-locus combinations from genome-wide datasets. However, for high-order combinations, existing methods either adopt a brute-force search which only handles a small number of SNPs (up to few hundreds), or use heuristic search that may miss informative combinations. In addition, existing approaches lack statistical power because of the use of statistics with high degrees-of-freedom and the huge number of hypotheses tested during combinatorial search. We designed an efficient and effective framework for high-order combinations in case-control datasets. The substantially improved efficiency and scalability demonstrated on synthetic and real datasets with several thousands of SNPs allows the study of several important mathematical and statistical properties of SNP combinations with order as high as eleven. We further explore functional interactions in high-order combinations and reveal a general connection between the increase in discriminative power of a combination over its subsets and the functional coherence among the genes comprising the combination, supported by multiple datasets. Finally, we study several significant high-order combinations discovered from a lung-cancer dataset and a kidney-transplant-rejection dataset in detail to provide novel insights on the complex diseases. Interestingly, many of these associations involve combinations of common variations that occur in small fractions of population. Thus, our approach is an alternative methodology for exploring the genetics of rare diseases for which the current focus is on individually rare variations.
PLoS ONE, 7(4): e33531. doi:10.1371/journal.pone.0033531, 2012 (*co-corresponding authors) (software)

Mining Low-support Discriminative Patterns from Dense and High-dimensional Data
Gang Fang, Gaurav Pandey, Wen Wang, Manish Gupta, Michael Steinbach and Vipin Kumar.

Discriminative patterns can provide valuable insights into data sets with class labels, that may not be available from the individual features or the predictive models built using them. Most existing approaches work efficiently for sparse or low-dimensional data sets. However, for dense and high-dimensional data sets, they have to use high thresholds to produce the complete results within limited time, and thus, may miss interesting low-support patterns. In this paper, we address the necessity of trading off the completeness of discriminative pattern discovery with the efficient discovery of low-support discriminative patterns from such data sets. We propose a family of antimonotonic measures named SupMaxK that organize the set of discriminative patterns into nested layers of subsets, which are progressively more complete in their coverage, but require increasingly more computation. In particular, the member of SupMaxK with K ¼ 2, named SupMaxPair, is suitable for dense and high-dimensional data sets. Experiments on both synthetic data sets and a cancer gene expression data set demonstrate that there are low-support patterns that can be discovered using SupMaxPair but not by existing approaches. Furthermore, we show that the low-support discriminative patterns that are only discovered using SupMaxPair from the cancer gene expression data set are statistically significant and biologically relevant. This illustrates the complementarity of SupMaxPair to existing approaches for discriminative pattern discovery.

IEEE Transaction on Knowledge and Data Engineering, vol 24(2), p 279-294, 2012 (software)