Integrating data sources

Clinical and molecular data are currently stored in many different databases using different semantics and different formats. A new project called DeepLink aims to develop a framework that would make it possible to compare and analyze data across platforms not originally intended to intersect. (Image courtesy of Nicholas Tatonetti.)

Medical doctors and basic biological scientists tend to speak about human health in different languages. Whereas doctors in the clinic focus on phenomena such as symptoms, drug effects, and treatment outcomes, basic scientists often concentrate on activity at the molecular and cellular levels such as genetic alterations, gene expression changes, or protein profiles. Although these various layers are all related physiologically, there is no standard terminology or framework for storing and organizing the different kinds of data that describe them, making it difficult for scientists to systematically integrate and analyze data across different biological scales. Being able to do so, many investigators now believe, could provide a more efficient and comprehensive way to understand and fight disease.

A new project recently launched by Nicholas Tatonetti (Assistant Professor in the Columbia University Departments of Systems Biology and Biomedical Informatics) along with co-principal investigators Chunhua Weng (Department of Biomedical Informatics) and Michel Dumontier (Stanford University), aims to bridge this divide. With the support of a $1.1 million grant from the National Center for Advancing Translational Science (NCATS) the scientists have begun to develop a tool they call DeepLink, a data translator that will integrate health-related findings at multiple scales.

As Dr. Tatonetti explains, “We want to close what we call the interoperability gap, a fundamental difference in the language and semantics used to describe the models and knowledge between the clinical and molecular domains. Our goal is to develop a scalable electronic architecture for integrating the enormous multiscale knowledge that is now available.”

Cell Types in Autism

By inventing a new computational pipeline called DAMAGES, Chaolin Zhang and Yufeng Shen showed that brain cell types on the left of the plot are more prone to have rare autism risk mutations than cell types at the right. Narrowing the focus to these types of cells also helped to identify a molecular signature of the disorder that involves haploinsufficiency. Figure: Human Mutation.

Autism, a spectrum of neurodevelopmental disorders typically identified during early childhood, is widely thought to be the result of genetic alterations that change how the growing brain is wired. Nevertheless, despite a substantial effort in the field of autism genetics, the specific alterations that place one child at greater risk than another remain elusive. Although the list of alterations associated with autism is growing, it has been difficult to conclusively distinguish those that truly increase disease risk from those that are merely coincident with it. One troubling reason for this is that research so far seems to indicate that specific genetic abnormalities associated with autism risk are extremely rare, with many being found only in single patients. This has made it hard to reproduce findings conclusively.

In a paper recently published in the journal Human Mutation, Department of Systems Biology faculty members Chaolin Zhang and Yufeng Shen describe a method and some new findings that could help to more precisely identify rare autism-driving alterations. A new analytical pipeline they call DAMAGES (Disease Associated Mutation Analysis using Gene Expression Signatures) uses a unique approach to identifying autism risk genes, looking at differences in gene expression among different cell types in the brain in order to focus more specifically on mechanisms that are likely to be relevant for autism. Using this approach, they identified a pronounced molecular signature that is shared by disease risk genes due to haploinsufficiency, a type of genetic alteration that causes a dramatic drop in the expression of a particular protein.

Yufeng Shen
Yufeng Shen's lab is interested in developing better computational methods for identifying rare genetic variants that increase disease risk.

On the surface, birth defects and cancer might not seem to have much in common. For some time, however, scientists have observed increased cancer risk among patients with certain developmental syndromes. One well-known example is seen in children with Noonan syndrome, who have an eightfold increased risk of developing leukemia. Recently, researchers studying the genetics of autism also observed mutations in PTEN, an important tumor suppressor gene. Although such findings have been largely isolated and anecdotal, they raise the tantalizing question of whether cancer and developmental disorders might be fundamentally linked.

According to a paper recently published in the journal Human Mutation, many of these similarities might not be just coincidental, but the result of shared genetic mutations. The study, led by Yufeng Shen, an Assistant Professor in the Columbia University Departments of Systems Biology and Biomedical Informatics, together with Wendy Chung, Kennedy Family Associate Professor of Pediatrics at Columbia University Medical Center, found that cancer-driving genes also make up more than a third of the risk genes for developmental disorders. Moreover, many of these genes appear to function through similar modes of action. The scientists suggest that this could make tumors “natural laboratories” for pinpointing and predicting the damaging effects of rare genetic alterations that cause developmental disorders.

“In comparison with cancer, there are relatively few patients with developmental disorders,” Shen explains, “For geneticists, this makes it hard to identify the risk genes solely based on statistical evidence of mutations from these patients. This study indicates that we should be able to use what we learn from cancer genetics — where much more data are available — to help in the interpretation of genetic data in developmental disorders.”

Clonal evolution in GBM tumors
The researchers' model of tumor evolution indicates that different clonal lineages branch from a common ancestral cell and then diversify, independently causing aggressive tumor behavior at different stages of disease.

Glioblastoma multiforme (GBM) is the most common and most aggressive type of primary brain tumor in adults. Existing treatments against the disease are very limited in their effectiveness, meaning that in most patients tumors recur within a year. Once GBM returns, no beneficial therapeutics currently exist and prognosis is generally very poor.

To better understand how GBM evades treatment, an international team led by Antonio Iavarone and Raul Rabadan at the Columbia University Center for Topology of Cancer Evolution and Heterogeneity has been studying how the cellular composition of GBM tumors changes over the course of therapy. In a paper just published online by Nature Genetics, they provide the first sketch of the main routes of GBM tumor evolution during treatment, showing that different cellular clones within a tumor become dominant within specific tumor states. The study uncovers important general principles of tumor evolution, novel genetic markers of disease progression, and new potential therapeutic targets.


By using statistical methods to compare genomic data across species, such as chimpanzees and humans, the Przeworski Lab is gaining insights into the origins of genetic variation and adaptation. (Photo: Common chimpanzee at the Leipzig Zoo. Thomas Lersch, Wikimedia Commons.)

Launched approximately 100 years ago, population genetics is a subfield within evolutionary biology that seeks to explain how processes such as mutation, natural selection, and random genetic drift lead to genetic variation within and between species. Population genetics was originally born from the convergence of Mendelian genetics and biostatistics, but with the recent availability of genome sequencing data and high-performance computing technologies, it has bloomed into a mature computational science that is providing increasingly high-resolution models of the processes that drive evolution.

Molly Przeworski, a professor in the Columbia University Departments of Biological Sciences and Systems Biology, majored in mathematics at Princeton before beginning her PhD in evolutionary biology at the University of Chicago in the mid-1990s. While there, she realized that the availability of increasingly large data sets was changing population genetics, and has since been interested in using statistical approaches to investigate questions such as how genetic variation drives adaptation and why mutation rate and recombination rate differ among species. In the following interview, she describes how population genetics is itself evolving, as well as some of her laboratory’s contributions to the field.

Yaniv Erlich
Yaniv Erlich. Photo: Jared Leeds.

A new article published online in Nature Genetics reports that short tandem repeats, a class of genetic alterations in which short motifs of nucleotide base pairs occur multiple times in a row, play a role in modulating gene expression. Leading the study was Yaniv Erlich, an assistant professor in the Columbia University Department of Computer Science and core member of the New York Genome Center who recently joined the Center for Computational Biology and Bioinformatics.

As an article in Columbia Engineering explains, the findings reveal a new class of genome regulation.

The Department of Systems Biology and Center for Computational Biology and Bioinformatics are pleased to announce that three Columbia University faculty members have recently joined our community. Kam Leong, the Samuel Y. Sheng Professor of Biomedical Engineering at Columbia University, is now an interdisciplinary faculty member in the Department of Systems Biology. In addition, Yaniv Erlich and Guy Sella are now members of the Center for Computational Biology and Bioinformatics (C2B2). Their addition to the Department and to C2B2 will bring new expertise that will benefit our research and education activities, incorporating perspectives from fields such as nanotechnology, bioinformatics, and evolutionary genomics.

Topology of cancer

The Columbia University Center for Topology of Cancer Evolution and Heterogeneity will combine mathematical approaches from topological data analysis with new single-cell experimental technologies to study cellular diversity in solid tumors. Image courtesy of Raul Rabadan.

The National Cancer Institute’s Physical Sciences in Oncology program has announced the creation of a new center for research and education based at Columbia University. The Center for Topology of Cancer Evolution and Heterogeneity will develop and utilize innovative mathematical and experimental techniques to explore how genetic diversity emerges in the cells that make up solid tumors. In this way it will address a key challenge facing cancer research in the age of precision medicine — how to identify the clonal variants within a tumor that are responsible for its growth, spread, and resistance to therapy. Ultimately, the strategies the Center develops could be used to identify more effective biomarkers of disease and new therapeutic strategies.

Comorbidity between Mendelian disease and cancer
Researchers in the Rabadan Lab have found that comorbidity between Mendelian diseases and cancer may result from shared genetic factors.

Genetic diseases can arise in a variety of ways. Mendelian disorders, for example, occur when specific mutations in single genes — called germline mutations — are inherited from either of one’s two parents. Well-known examples of Mendelian diseases include cystic fibrosis, sickle cell disease, and Duchenne muscular dystrophy. Other genetic diseases, including cancer, result from somatic mutations, which occur in individual cells during a person’s lifetime. Because the genetic origins of Mendelian diseases and cancer are so different, they are typically understood to be distinct phenomena. However, scientists in the Columbia University Department of Systems Biology have found evidence that there might be interesting genetic connections between them. 

In a paper just published in Nature Communications, postdoctoral research scientist Rachel Melamed and colleagues in the laboratory of Associate Professor Raul Rabadan report on a new method that uses knowledge about Mendelian diseases to suggest mutations involved in cancer. The study takes advantage of an enormous collection of electronic health records representing over 110 million patients, a substantial percentage of US residents. The authors show that clinical co-occurrence of Mendelian diseases and cancer, known as comorbidity, can be tied to genetic changes that play roles in both diseases. The paper also identifies several specific relationships between Mendelian diseases and the cancers melanoma and glioblastoma.

ALK-negative ALCL mutation map
A map of mutations observed in ALK-negative anaplastic large cell lymphoma. (Credit: Dr. Rabadan)

The following article is reposted with permission from the Columbia University Medical Center Newsroom. Find the original here.

The first-ever systematic study of the genomes of patients with ALK-negative anaplastic large cell lymphoma (ALCL), a particularly aggressive form of non-Hodgkin’s lymphoma (NHL), shows that many cases of the disease are driven by alterations in the JAK/STAT3 cell signaling pathway. The study also demonstrates, in mice implanted with human-derived ALCL tumors, that the disease can be inhibited by compounds that target this pathway, raising hopes that more effective treatments might soon be developed. The study, led by researchers at Columbia University Medical Center (CUMC) and Weill Cornell Medical College, was published today in the online edition of Cancer Cell.

Autism Spectrum Disorders Genetic Network

Network of autism-associated genes. (Credit: Dennis Vitkup)

The following article is reposted with permission from the Columbia University Medical Center Newsroom. Find the original here.

People with autism have a wide range of symptoms, with no two people sharing the exact type and severity of behaviors. Now a large-scale analysis of hundreds of patients and nearly 1000 genes has started to uncover how diversity among traits can be traced to differences in patients’ genetic mutations. The study, conducted by researchers at Columbia University Medical Center, was published Dec. 22 in the journal Nature Neuroscience.

Autism researchers have identified hundreds of genes that, when mutated, likely increase the risk of developing autism spectrum disorder (ASD). Much of the variability among people with ASD is thought to stem from the diversity of underlying genetic changes, including the specific genes mutated and the severity of the mutation.

“If we can understand how different mutations lead to different features of ASD, we may be able to use patients’ genetic profiles to develop accurate diagnostic and prognostic tools and perhaps personalize treatment,” said senior author Dennis Vitkup, PhD, associate professor of systems biology and biomedical informatics at Columbia University’s College of Physicians & Surgeons.

Sequence of genomic alterations in CLLA graph representing the sequence of genomic alterations in chronic lymphocytic leukemia (CLL). Each node represents a mutation, with arrows indicating temporal relationships between them. The size of the nodes indicates the number of patients in the study who exhibited the alteration, while the thickness of the lines shows how often the temporal relationships between nodes were seen. The method the researchers use enabled them to identify multiple, distinct evolutionary patterns in CLL.

As biologists have gained a better understanding of cancer, it has become clear that tumors are often driven not by a single mutation, but by a series of genetic changes that correspond to particular stages of cancer progression. In this sense, a tumor is constantly evolving, with different groups of cells that harbor distinctive mutations multiplying at different rates, depending on their fitness for particular disease states. As the search for more effective cancer diagnostics and therapies continues, one key question is how to disentangle the order in which mutations occur in order to understand how tumors change over time. Being able to predict how a tumor will behave based on signs seen early in the course of disease could enable the development of new diagnostics that could better inform treatment planning.

In a paper just published in the journal eLife, a team of investigators led by Department of Systems Biology Associate Professor Raul Rabadan reports on a new computational strategy for addressing this challenge. Their framework, called tumor evolutionary directed graphs (TEDG), considers next-generation sequencing data from tumor samples from a large number of patients. Using TEDG to analyze cancer cells in patients with chronic lymphocytic leukemia (CLL), they were able to develop a model of how the disease’s mutational landscape changes from its initial onset to its late stages. Their findings suggest that CLL may not be just the result of a single evolutionary path, but can evolve in alternative ways.

Expanding the landscape of breast cancer drivers

In comparison with a previous study (Stephens et al., 2012, shown in gray), a new computational approach that focuses on somatic copy number mutations increased the number of known driver mutations in breast tumors to a median of five for each tumor. The findings could raise the likelihood of finding actionable targets in individual patients with breast cancer.

For many years, researchers have known that somatic copy number alterations (SCNA’s) — insertions, deletions, duplications, and transpositions of sections of DNA that are not inherited but occur after birth — play important roles in causing many types of cancer. Indeed, most recurrent drivers of epithelial tumors are copy number alterations, with some found in up to 40% of patients with specific tumor types. However, because SCNA’s occur when entire sections of chromosomes become damaged, biologists have had difficulty developing effective methods for distinguishing genes within SCNA’s that actually drive cancer from those genes that might lie near a driver but do not themselves cause disease.

Helios nearly doubled the number of high-confidence predictions of breast cancer drivers.

In a new paper published in Cell, researchers in the laboratories of Dana Pe’er (Columbia University Departments of Systems Biology and Biological Sciences) and Jose Silva (Icahn School of Medicine at Mount Sinai) report on a new computational algorithm that promises to dramatically improve researchers’ ability to identify cancer-driving genes within potentially large SCNA’s. The algorithm, called Helios, was used to analyze a combination of genomic data and information generated by functional RNAi screens, enabling them to predict several dozen new SCNA drivers of breast cancer. In follow-up in vitro experimental studies, they tested 12 of these predictions, 10 of which were validated in the laboratory. Their findings nearly double the number of breast cancer drivers, providing many new opportunities towards personalized treatments for breast cancer. Their methodology is general and could also be used to locate disease-causing SCNA’s in other cancer types.

Leading this effort was Felix Sanchez-Garcia, a recent PhD graduate from the Pe’er Lab and a first author on the paper. The story of how this breakthrough came about illuminates how the interdisciplinary research and education that take place at the Department of Systems Biology can address important challenges facing biological and biomedical research.

DIGGIT identifies mutations upstream of master regulators.

A new algorithm called DIGGIT identifies mutations that lie upstream of crucial bottlenecks within regulatory networks. These bottlenecks, called master regulators, integrate these mutations and become essential functional drivers of diseases such as cancer.

Although genome-wide association studies have made it possible to identify mutations that are linked to diseases such as cancer, determining which mutations actually drive disease and the mechanics of how they do so has been an ongoing challenge. In a paper just published in Cell, researchers in the lab of Andrea Califano describe a new computational approach that may help address this problem.

Ashkenazi Population Bottleneck Model
The consortium’s model of Ashkenazi Jewish ancestry suggests that the population’s history was shaped by three critical bottleneck events. The ancestors of both populations underwent a bottleneck sometime between 85,000 and 91,000 years ago, which was likely coincident with an Out-of-Africa event. The founding European population underwent a bottleneck at approximately 21,000 years ago, beginning a period of interbreeding between individuals of European and Middle Eastern ancestry. A severe bottleneck occurred in the Middle Ages, reducing the population to under 350 individuals. The modern-day Ashkenazi community emerged from this group.

An international research consortium led by Associate Professor Itsik Pe’er has produced a new panel of reference genomes that will significantly improve the study of genetic variation in Ashkenazi Jews. Using deep sequencing to analyze the genomes of 128 healthy individuals of Ashkenazi Jewish origin, The Ashkenazi Genome Consortium (TAGC) has just published a resource that will be much more effective than previously available European reference genomes for identifying disease-causing mutations within this historically isolated population. Their study also provides novel insights into the historical origins and ancestry of the Ashkenazi community. A paper describing their study has just been published online in Nature Communications.

The dataset produced by the consortium provides a high-resolution baseline genomic profile of the Ashkenazi Jewish population, which they revealed to be significantly different from that found in non-Jewish Europeans. In the past, clinicians’ only option for identifying disease-causing mutations in Ashkenazi individuals was to compare their genomes to more heterogeneous European reference sets. This new resource accounts for the historical isolation of this population, and so will make genetic screening much more accurate in identifying disease-causing mutations.

In an article that appears on the website of Columbia University’s Fu Foundation School of Engineering and Computer Science, Dr. Pe’er explains:

“Our study is the first full DNA sequence dataset available for Ashkenazi Jewish genomes... With this comprehensive catalog of mutations present in the Ashkenazi Jewish population, we will be able to more effectively map disease genes onto the genome and thus gain a better understanding of common disorders. We see this study serving as a vehicle for personalized medicine and a model for researchers working with other populations.”

In addition to offering an important resource for such future translational and clinical research, the paper’s findings also provide new insights that have implications for the much debated question of how European and Ashkenazi Jewish populations emerged historically.

Comparing human and mouse prostate cancer networks

Computational synergy analysis depicting FOXM1 and CENPF regulons from the human (left) and mouse (right) interactomes showing shared and nonshared targets. Red corresponds to overexpressed targets and blue to underexpressed targets.

Two genes work together to drive the most lethal forms of prostate cancer, according to new research by investigators in the Columbia University Department of Systems Biology.  These findings could lead to a diagnostic test for identifying those tumors likely to become aggressive and to the development of novel combination therapy for the disease.

The two genes—FOXM1 and CENPF—had been previously implicated in cancer, but none of the prior studies suggested that they might work synergistically to cause the most aggressive form of prostate cancer. The study was published today in the online issue of Cancer Cell.

Molly PrzeworskiMolly Przeworski has joined Columbia University as Professor in the Department of Systems Biology and Department of Biological Sciences. The Przeworski lab investigates how natural selection, genetic drift, mutation, and recombination shape the heritable differences seen among individuals and species. To this end, they develop models of the evolutionary process, create statistical tools, and analyze large-scale variation data sets. Among the goals of their research are to understand how natural selection has shaped patterns of genetic variation, and to identify the causes and consequences of variation in recombination and mutation rates, in humans and other organisms.

Tuuli LappalainenTuuli Lappalainen has joined Columbia University as an assistant professor in the Department of Systems Biology. Dr. Lappalainen is a specialist in the analysis of RNA sequencing data, with research interests including functional variation in the human genome, population genetic background of variation in the human genome, and interpretation of genome function.

Dr. Lappalainen joins the Department of Systems Biology in co-appointment with the New York Genome Center (NYGC), where she will also serve as a Junior Investigator and Core Member. Based in lower Manhattan, NYGC is a consortium made up primarily of New York-area institutions that is designed to translate promising genomics-based research into new strategies for treating, preventing, and managing disease. This co-appointment with Columbia University — an institutional founding member of the NYGC — will enhance collaboration between the two institutions. (Read an interview with Dr. Lappalainen at the New York Genome Center website.)

Reversing glucocorticoid resistance

A representative example of tumor load analysis using bioluminescence imaging in mice following xenograft with T-ALL. Treatment with either MK2206 or dexamethasone showed limited efficacy, while combination treatment saw near complete elimination of tumor cells.

In a paper published in Cancer Cell, a team of researchers led by Adolfo Ferrando and Andrea Califano at Columbia University has identified the protein kinase AKT as a target for reversing resistance to glucocorticoid therapy in patients with acute lymphoblastic leukemia (ALL).  

Researchers in the Columbia University Department of Systems Biology and Herbert Irving Comprehensive Cancer Center have determined that measuring the expression levels of three genes associated with aging can be used to predict the aggressiveness of seemingly low-risk prostate cancer. Use of this three-gene biomarker, in conjunction with existing cancer-staging tests, could help physicians better determine which men with early prostate cancer can be safely followed with “active surveillance” and spared the risks of prostate removal or other invasive treatment. The findings were published today in the online edition of Science Translational Medicine.

More than 200,000 new cases of prostate cancer are diagnosed each year in the U.S. “Most of these cancers are slow growing and will remain so, and thus they do not require treatment,” said study leader Cory Abate-Shen, Michael and Stella Chernow Professor of Urological Oncology at Columbia University Medical Center (CUMC). “The problem is that, with existing tests, we cannot identify the small percentage of slow-growing tumors that will eventually become aggressive and spread beyond the prostate. The three-gene biomarker could take much of the guesswork out of the diagnostic process and ensure that patients are neither overtreated nor undertreated.”