Developing Bioinformatics Computer Skills: An Introduction to Software Tools for Biological Applications

by Per Jambeck, Cynthia Gibas

Overview

Bioinformatics—the application of computational and analytical methods to biological problems—is a rapidly evolving scientific discipline. Genome sequencing projects are producing vast amounts of biological data for many different organisms, and, increasingly, storing these data in public databases. Such biological databases are growing exponentially, along with the biological literature. It's impossible for even the most zealous researcher to stay on top of necessary information in the field without the aid of computer-based tools. Bioinformatics is all about building these tools. Developing Bioinformatics Computer Skills is for scientists and students who are learning computational approaches to biology for the first time, as well as for experienced biology researchers who are just starting to use computers to handle their data. The book covers the Unix file system, building tools and databases for bioinformatics, computational approaches to biological problems, an introduction to Perl for bioinformatics, data mining, and data visualization. Written in a clear, engaging style, Developing Bioinformatics Computer Skills will help biologists develop a structured approach to biological data as well as the tools they'll need to analyze the data.

Product Details

ISBN-13:	9781565926646
Publisher:	O'Reilly Media, Incorporated
Publication date:	05/28/2001
Pages:	446
Product dimensions:	7.00(w) x 9.19(h) x 0.95(d)

About the Author

Per Jambeck is a Ph.D. student in the bioengineering department at the University of California, San Diego. He has worked on computational biology problems since 1994, concentrating on machine learning applications in understanding multidimensional biological data. Per smiles wistfully at the mention of free time, but he manages to host shows at community and student-run radio stations anyway.

Cynthia Gibas is an assistant professor of biology at Virginia Tech, in Blacksburg, Virginia. She's been a computational biologist since before computational biology was cool, and is currently learning to drive her spankin' new home-built Linux cluster. Her research interests include the structure and evolution of genomes, the properties of protein surfaces and interfaces, and prediction of protein structure. She teaches introductory courses in bioinformatics methods for biologists and is looking forward to her next real vacation, sometime in 2006.

Read an Excerpt

Chapter 1: Biology in the Computer Age

From the interaction of species and populations, to the function of tissues and cells within an individual organism, biology is defined as the study of living things. In the course of that study, biologists collect and interpret data. Now, at the beginning of the 21st century, we use sophisticated laboratory technology that allows us to collect data faster than we can interpret it. We have vast volumes of DNA sequence data at our fingertips. But how do we figure out which parts of that DNA control the various chemical processes of life? We know the function and structure of some proteins, but how do we determine the function of new proteins? And how do we predict what a protein will look like, based on knowledge of its sequence? We understand the relatively simple code that translates DNA into protein. But how do we find meaningful new words in the code and add them to the DNA-protein dictionary?

Bioinformatics is the science of using information to understand biology; it's the tool we can use to help us answer these questions and many others like them. Unfortunately, with all the hype about mapping the human genome, bioinformatics has achieved buzzword status; the term is being used in a number of ways, depending on who is using it. Strictly speaking, bioinformatics is a subset of the larger field of computational biology , the application of quantitative analytical techniques in modeling biological systems. In this book, we stray from bioinformatics into computational biology and back again. The distinctions between the two aren't important for our purpose here, which is to cover a range of tools and techniques we believe are critical for molecular biologists who want to understand and apply the basic computational tools that are available today.

The field of bioinformatics relies heavily on work by experts in statistical methods and pattern recognition. Researchers come to bioinformatics from many fields, including mathematics, computer science, and linguistics. Unfortunately, biology is a science of the specific as well as the general. Bioinformatics is full of pitfalls for those who look for patterns and make predictions without a complete understanding of where biological data comes from and what it means. By providing algorithms, databases, user interfaces, and statistical tools, bioinformatics makes it possible to do exciting things such as compare DNA sequences and generate results that are potentially significant. "Potentially significant" is perhaps the most important phrase. These new tools also give you the opportunity to overinterpret data and assign meaning where none really exists. We can't overstate the importance of understanding the limitations of these tools. But once you gain that understanding and become an intelligent consumer of bioinformatics methods, the speed at which your research progresses can be truly amazing.

How Is Computing Changing Biology?

An organism's hereditary and functional information is stored as DNA, RNA, and proteins, all of which are linear chains composed of smaller molecules. These macromolecules are assembled from a fixed alphabet of well-understood chemicals: DNA is made up of four deoxyribonucleotides (adenine, thymine, cytosine, and guanine), RNA is made up from the four ribonucleotides (adenine, uracil, cytosine, and guanine), and proteins are made from the 20 amino acids. Because these macromolecules are linear chains of defined components, they can be represented as sequences of symbols. These sequences can then be compared to find similarities that suggest the molecules are related by form or function.

Sequence comparison is possibly the most useful computational tool to emerge for molecular biologists. The World Wide Web has made it possible for a single public database of genome sequence data to provide services through a uniform interface to a worldwide community of users. With a commonly used computer program called fsBLAST, a molecular biologist can compare an uncharacterized DNA sequence to the entire publicly held collection of DNA sequences. In the next section, we present an example of how sequence comparison using the BLAST program can help you gain insight into a real disease.

The Eye of the Fly

Fruit flies (Drosophila melanogaster) are a popular model system for the study of development of animals from embryo to adult. Fruit flies have a gene called eyeless, which, if it's "knocked out (i.e., eliminated from the genome using molecular biology methods), results in fruit flies with no eyes. It's obvious that the eyeless gene plays a role in eye development.

Researchers have identified a human gene responsible for a condition called aniridia. In humans who are missing this gene (or in whom the gene has mutated just enough for its protein product to stop functioning properly), the eyes develop without irises.

If the gene for aniridia is inserted into an eyeless drosophila "knock out," it causes the production of normal drosophila eyes. It's an interesting coincidence. Could there be some similarity in how eyeless and aniridia function, even though flies and humans are vastly different organisms? Possibly. To gain insight into how eyeless and aniridia work together, we can compare their sequences. Always bear in mind, however, that genes have complex effects on one another. Careful experimentation is required to get a more definitive answer.

As little as 15 years ago, looking for similarities between eyeless and aniridia DNA sequences would have been like looking for a needle in a haystack. Most scientists compared the respective gene sequences by hand-aligning them one under the other in a word processor and looking for matches character by character. This was time-consuming, not to mention hard on the eyes.

In the late 1980s, fast computer programs for comparing sequences changed molecular biology forever. Pairwise comparison of biological sequences is the foundation of most widely used bioinformatics techniques. Many tools that are widely available to the biology community--including everything from multiple alignment, phylogenetic analysis, motif identification, and homology-modeling software, to web-based database search services--rely on pairwise sequence-comparison algorithms as a core element of their function.

These days, a biologist can find dozens of sequence matches in seconds using sequence-alignment programs such as BLAST and FASTA. These programs are so commonly used that the first encounter you have with bioinformatics tools and biological databases will probably be through the National Center for Biotechnology Information's (NCBI) BLAST web interface. Figure 1-1 shows a standard form for submitting data to NCBI for a BLAST search.

Labels in Gene Sequences Before you rush off to compare the sequences of eyeless and aniridia with BLAST, let us tell you a little bit about how sequence alignment works.

It's important to remember that biological sequence (DNA or protein) has a chemical function, but when it's reduced to a single-letter code, it also functions as a unique label, almost like a bar code. From the information technology point of view, sequence information is priceless. The sequence label can be applied to a gene, its product, its function, its role in cellular metabolism, and so on. The user searching for information related to a particular gene can then use rapid pairwise sequence comparison to access any information that's been linked to that sequence label.

The most important thing about these sequence labels, though, is that they don't just uniquely identify a particular gene; they also contain biologically meaningful patterns that allow users to compare different labels, connect information, and make inferences. So not only can the labels connect all the information about one gene, they can help users connect information about genes that are slightly or even dramatically different in sequence.

If simple labels were all that was needed to make sense of biological data, you could just slap a unique number (e.g., a GenBank ID) onto every DNA sequence and be done with it. But biological sequences are related by evolution, so a partial pattern match between two sequence labels is a significant find. BLAST differs from simple keyword searching in its ability to detect partial matches along the entire length of a protein sequence...

Isn't Bioinformatics Just About Building Databases?

Much of what we currently think of as part of bioinformatics--sequence comparison, sequence database searching, sequence analysis--is more complicated than just designing and populating databases. Bioinformaticians (or computational biologists) go beyond just capturing, managing, and presenting data, drawing inspiration from a wide variety of quantitative fields, including statistics, physics, computer science, and engineering. Figure 1-2 shows how quantitative science intersects with biology at every level, from analysis of sequence data and protein structure, to metabolic modeling, to quantitative analysis of populations and ecology.

Bioinformatics is first and foremost a component of the biological sciences. The main goal of bioinformatics isn't developing the most elegant algorithms or the most arcane analyses; the goal is finding out how living things work. Like the molecular biology methods that greatly expanded what biologists were capable of studying, bioinformatics is a tool and not an end in itself. Bioinformaticians are the tool-builders, and it's critical that they understand biological problems as well as computational solutions in order to produce useful tools.

Research in bioinformatics and computational biology can encompass anything from abstraction of the properties of a biological system into a mathematical or physical model, to implementation of new algorithms for data analysis, to the development of databases and web tools to access them.

The First Information Age in Biology Biology as a science of the specific means that biologists need to remember a lot of details as well as general principles. Biologists have been dealing with problems of information management since the 17th century.

The roots of the concept of evolution lie in the work of early biologists who catalogued and compared species of living things. The cataloguing of species was the preoccupation of biologists for nearly three centuries, beginning with animals and plants and continuing with microscopic life upon the invention of the compound microscope. New forms of life and fossils of previously unknown, extinct life forms are still being discovered even today.

All this cataloguing of plants and animals resulted in what seemed a vast amount of information at the time. In the mid-16th century, Otto Brunfels published the first major modern work describing plant species, the Herbarium vitae eicones. As Europeans traveled more widely around the world, the number of catalogued species increased, and botanical gardens and herbaria were established. The number of catalogued plant types was 500 at the time of Theophrastus, a student of Aristotle. By 1623, Casper Bauhin had observed 6,000 types of plants. Not long after John Ray introduced the concept of distinct species of animals and plants, and developed guidelines based on anatomical features for distinguishing conclusively between species. In the 1730s, Carolus Linnaeus catalogued 18,000 plant species and over 4,000 species of animals, and established the basis for the modern taxonomic naming system of kingdoms, classes, genera, and species. By the end of the 18th century, Baron Cuvier had listed over 50,000 species of plants.

It was no coincidence that a concurrent preoccupation of biologists, at this time of exploration and cataloguing, was classification of species into an orderly taxonomy. A botany text might encompass several volumes of data, in the form of painstaking illustrations and descriptions of each species encountered. Biologists were faced with the problem of how to organize, access, and sensibly add to this information. It was apparent to the casual observer that some living things were more closely related than others. A rat and a mouse were clearly more similar to each other than a mouse and a dog. But how would a biologist know that a rat was like a mouse (but that rat was not just another name for mouse) without carrying around his several volumes of drawings? A nomenclature that uniquely identified each living thing and summed up its presumed relationship with other living things, all in a few words, needed to be invented.

The solution was relatively simple, but at the time, a great innovation. Species were to be named with a series of one-word names of increasing specificity. First a very general division was specified: animal or plant? This was the kingdom to which the organism belonged. Then, with increasing specificity, came the names for class, genera, and species... A modern taxonomy of the earth's millions of species is too complicated for even the most zealous biologist to memorize, and fortunately computers now provide a way to maintain and access the taxonomy of species. The University of Arizona's Tree of Life project and NCBI's Taxonomy database are two examples of online taxonomy projects.

Taxonomy was the first informatics problem in biology. Now, biologists have reached a similar point of information overload by collecting and cataloguing information about individual genes. The problem of organizing this information and sharing knowledge with the scientific community at the gene level isn't being tackled by developing a nomenclature. It's being attacked directly with computers and databases from the start.

The evolution of computers over the last half-century has fortuitously paralleled the developments in the physical sciences that allow us to see biological systems in increasingly fine detail...

Simply finding the right needles in the haystack of information that is now available can be a research problem in itself. Even in the late 1980s, finding a match in a sequence database was worth a five-page publication. Now this procedure is routine, but there are many other questions that follow on our ability to search sequence and structure databases. These questions are the impetus for the field of bioinformatics.

What Does Informatics Mean to Biologists?

The science of informatics is concerned with the representation, organization, manipulation, distribution, maintenance, and use of information, particularly in digital form. There is more than one interpretation of what bioinformatics--the intersection of informatics and biology--actually means, and it's quite possible to go out and apply for a job doing bioinformatics and find that the expectations of the job are entirely different than you thought.

The functional aspect of bioinformatics is the representation, storage, and distribution of data. Intelligent design of data formats and databases, creation of tools to query those databases, and development of user interfaces that bring together different tools to allow the user to ask complex questions about the data are all aspects of the development of bioinformatics infrastructure.

Developing analytical tools to discover knowledge in data is the second, and more scientific, aspect of bioinformatics. There are many levels at which we use biological information, whether we are comparing sequences to develop a hypothesis about the function of a newly discovered gene, breaking down known 3D protein structures into bits to find patterns that can help predict how the protein folds, or modeling how proteins and metabolites in a cell work together to make the cell function. The ultimate goal of analytical bioinformaticians is to develop predictive methods that allow scientists to model the function and phenotype of an organism based only on its genome sequence. This is a grand goal, and one that will be approached only in small steps, by many scientists working together.

What Challenges Does Biology Offer Computer Scientists?

The goal of biology, in the era of the genome projects, is to develop a quantitative understanding of how living things are built from the genome that encodes them.

Cracking the genome code is complex. At the very simplest level, we still have difficulty identifying unknown genes by computer analysis of genomic sequence. We still have not managed to predict or model how a chain of amino acids folds into the specific structure of a functional protein.

Beyond the single-molecule level, the challenges are immense. The sheer amount of data in GenBank is now growing at an exponential rate, and as datatypes beyond DNA, RNA, and protein sequence begin to undergo the same kind of explosion, simply managing, accessing, and presenting this data to users in an intelligible form is a critical task. Human-computer interaction specialists need to work closely with academic and clinical researchers in the biological sciences to manage such staggering amounts of data.

Biological data is very complex and interlinked. A spot on a DNA array, for instance, is connected not only to immediate information about its intensity, but to layers of information about genomic location, DNA sequence, structure, function, and more. Creating information systems that allow biologists to seamlessly follow these links without getting lost in a sea of information is also a huge opportunity for computer scientists.

Finally, each gene in the genome isn't an independent entity. Multiple genes interact to form biochemical pathways, which in turn feed into other pathways. Biochemistry is influenced by the external environment, by interaction with pathogens, and by other stimuli. Putting genomic and biochemical data together into quantitative and predictive models of biochemistry and physiology will be the work of a generation of computational biologists. Computer scientists, mathematicians, and statisticians will be a vital part of this effort.

What Skills Should a Bioinformatician Have?

There's a wide range of topics that are useful if you're interested in pursuing bioinformatics, and it's not possible to learn them all. However, in our conversations with scientists working at companies such as Celera Genomics and Eli Lilly, we've picked up on the following "core requirements" for bioinformaticians:

• You should have a fairly deep background in some aspect of molecular biology. It can be biochemistry, molecular biology, molecular biophysics, or even molecular modeling, but without a core of knowledge of molecular biology you will, as one person told us, "run into brick walls too often."

• You must absolutely understand the central dogma of molecular biology. Understanding how and why DNA sequence is transcribed into RNA and translated into protein is vital. (In Chapter 2, Computational Approaches to Biological Questions, we define the central dogma, as well as review the processes of transcription and translation.)

• You should have substantial experience with at least one or two major molecular biology software packages, either for sequence analysis or molecular modeling. The experience of learning one of these packages makes it much easier to learn to use other software quickly.

• You should be comfortable working in a command-line computing environment. Working in Linux or Unix will provide this experience.

• You should have experience with programming in a computer language such as C/C++, as well as in a scripting language such as Perl or Python.

There are a variety of other advanced skill sets that can add value to this background: molecular evolution and systematics; physical chemistry--kinetics, thermodynamics and statistical mechanics; statistics and probabilistic methods; database design and implementation; algorithm development; molecular biology laboratory methods; and others.

Why Should Biologists Use Computers?

Computers are powerful devices for understanding any system that can be described in a mathematical way. As our understanding of biological processes has grown and deepened, it isn't surprising, then, that the disciplines of computational biology and, more recently, bioinformatics, have evolved from the intersection of classical biology, mathematics, and computer science.

A New Approach to Data Collection Biochemistry is often an anecdotal science. If you notice a disease or trait of interest, the imperative to understand it may drive the progress of research in that direction. Based on their interest in a particular biochemical process, biochemists have determined the sequence or structure or analyzed the expression characteristics of a single gene product at a time. Often this leads to a detailed understanding of one biochemical pathway or even one protein. How a pathway or protein interacts with other biological components can easily remain a mystery, due to lack of hands to do the work, or even because the need to do a particular experiment isn't communicated to other scientists effectively.

The Internet has changed how scientists share data and made it possible for one central warehouse of information to serve an entire research community. But more importantly, experimental technologies are rapidly advancing to the point at which it's possible to imagine systematically collecting all the data of a particular type in a central "factory" and then distributing it to researchers to be interpreted.

In the 1990s, the biology community embarked on an unprecedented project: sequencing all the DNA in the human genome. Even though a first draft of the human genome sequence has been completed, automated sequencers are still running around the clock, determining the entire sequences of genomes from various life forms that are commonly used for biological research. And we're still fine-tuning the data we've gathered about the human genome over the last 10 years. Immense strings of data, in which the locations of only a relatively few important genes are known, have been and still are being generated. Using image-processing techniques, maps of entire genomes can now be generated much more quickly than they could with chemical mapping techniques, but even with this technology, complete and detailed mapping of the genomic data that is now being produced may take years.

Recently, the techniques of x-ray crystallography have been refined to a degree that allows a complete set of crystallographic reflections for a protein to be obtained in minutes instead of hours or days. Automated analysis software allows structure determination to be completed in days or weeks, rather than in months. It has suddenly become possible to conceive of the same type of high-throughput approach to structure determination that the Human Genome Project takes to sequence determination. While crystallization of proteins is still the limiting step, it's likely that the number of protein structures available for study will increase by an order of magnitude within the next 5 to 10 years.

Parallel computing is a concept that has been around for a long time. Break a problem down into computationally tractable components, and instead of solving them one at a time, employ multiple processors to solve each subproblem simultaneously. The parallel approach is now making its way into experimental molecular biology with technologies such as the DNA microarray. Microarray technology allows researchers to conduct thousands of gene expression experiments simultaneously on a tiny chip. Miniaturized parallel experiments absolutely require computer support for data collection and analysis. They also require the electronic publication of data, because information in large datasets that may be tangential to the purpose of the data collector can be extremely interesting to someone else. Finding information by searching such databases can save scientists literally years of work at the lab bench.

The output of all these high-throughput experimental efforts can be shared only because of the development of the World Wide Web and the advances in communication and information transfer that the Web has made possible.

The increasing automation of experimental molecular biology and the application of information technology in the biological sciences have lead to a fundamental change in the way biological research is done. In addition to anecdotal research--locating and studying in detail a single gene at a time--we are now cataloguing all the data that is available, making complete maps to which we can later return and mark the points of interest. This is happening in the domains of sequence and structure, and has begun to be the approach to other types of data as well. The trend is toward storage of raw biological data of all types in public databases, with open access by the research community. Instead of doing preliminary research in the lab, scientists are going to the databases first to save time and resources...

Table of Contents

• Preface
• Introduction
  • Chapter 1: Biology in the Computer Age
  • Chapter 2: Computational Approaches to Biological Questions
• The Bioinformatics Workstation
  • Chapter 3: Setting Up Your Workstation
  • Chapter 4: Files and Directories in Unix
  • Chapter 5: Working on a Unix System
• Tools for Bioinformatics
  • Chapter 6: Biological Research on the Web
  • Chapter 7: Sequence Analysis, Pairwise Alignment, and Database Searching
  • Chapter 8: Multiple Sequence Alignments, Trees, and Profiles
  • Chapter 9: Visualizing Protein Structures and Computing Structural Properties
  • Chapter 10: Predicting Protein Structure and Function from Sequence
  • Chapter 11: Tools for Genomics and Proteomics
• Databases and Visualization
  • Chapter 12: Automating Data Analysis with Perl
  • Chapter 13: Building Biological Databases
  • Chapter 14: Visualization and Data Mining
• Bibliography
• Colophon

Editorial Reviews

Examining the application of computational and analytical methods to biological problems, this guide for biologists, researchers, and students, offers a structured approach to biological data and computer tools for analyzing it. The book covers the Unix file system, the development of tools and databases, computational approaches to biological problems, the use of Perl, data mining, data visualization, and customizing data analysis software. Gibas teaches biology at Virginia Tech; Jambeck studies bioengeering at the University of California, San Diego. Annotation c. Book News, Inc., Portland, OR (booknews.com)