Principles of Plant Genetics and Breeding

240 CHAPTER 14
Introduction
Bioinformatics is the application of informatic techniques to biological data. These techniques include acquiring, annotating,
analyzing, and archiving the biological data, using concepts from biology, computer science, and mathematics. Bioinformatics
has its roots in the work of people like Margaret Dayhoff (collecting known sequences (Dayhoff et al. 1965) and a mathematical
model of protein evolution (Dayhoff et al. 1978)) and David Sankoff (sequence alignment and statistical tests for homology) in
molecular phylogenetics.
Bioinformatic analyses aim to discover precise, testable hypotheses to supplement or redirect biology experiments. The results
of these bioinformatics-derived experiments then influence the next round of computer-based studies. This interaction between
experiment and computation speeds scientific progress.
The first large collections of biological data were protein sequences, followed by nucleic acid sequences, and were collected
by individual laboratories and stored on computer punch cards, along with annotations, such as species, biochemical function,
and physiological role, and functional and structural domains. As the amount of sequence data grew and the need to share this
data between multiple laboratories grew, the collection and curation were taken over by specialized groups such as National
Biomedical Research Foundation/Protein Information Resource (NBRF/PIR), GeneBank, EMBL, and Swiss-Prot, where curation
included standardizing the format and ancillary data. With increased size also came the need for tools to search, analyze, and
annotate these databases.
Pairwise alignment and database searching
Molecular biologists commonly isolate and sequence molecules based on their association with particular biological phenomena,
such as disease resistance in plants. Typically, the biochemical function of the newly determined sequence is not known and
one compares the newly determined sequence to all known sequences whose biochemical functions are known to generate a
testable hypothesis about its function. Thus, an early bioinformatics tool was to search a database of annotated sequences with a
newly determined sequence to find all similar sequences. Sequences that were similar enough were inferred to be homologous,
that is to have descended from a common evolutionary ancestor. The inference of homology generates the hypothesis that the
two molecules carry out the same biochemical function and perhaps the same physiological role. A complete discussion of
database searching is given in Nicholas et al. (2000).
The power of a successful database search is demonstrated by comparing the histories of cystic fibrosis (CF) and type I
neurofibromatosis (NF-1) research. Both disease genes were isolated in 1988. The CF gene was identified as a chloride ion
transport protein, which led to the development of a number of therapies, with many now in final clinical trials. In 1988
the database search with the NF-1 gene failed and no homologues were found. It was not until 1998 that NF-1 was identified
as a growth suppressor, which has rapidly lead to improved diagnosis and many potential therapies for which clinical trials
are just beginning. Thus, the successful identification of the CF gene as a chloride ion transporter accelerated this research
area by a decade compared to the time required to discover the biochemical function of the NF-1 gene through biological
experiments.
Multiple sequence alignment
Industry highlights
Bioinformatics for sequence and genomic data
Hugh B. Nicholas, Jr., David W. Deerfield II, and Alexander J. Ropelewski
Pittsburgh Supercomputing Center, Pittsburgh, PA 15213, USA
A database search results in discovering many similar sequences from which one would like to create a multiple sequence alignment
that simultaneously shows the relationship among its homologous residues in the other sequences. This alignment is a map
of the evolution of the protein family. The multiple sequence alignment is a rich source of hypotheses to guide experimental work
since the alignment contains patterns of conservation and variation of residues among the sequences, which provides insights
into functional and structural positions for either the family of proteins or the genes encoding them. Such inferences are strongest
if the alignment contains sequences from widely diverse species.
Multiple sequence alignment implies that the residues in each column of the alignment are all evolutionarily related to each
other. Thus accuracy is most commonly considered to be improved by maximizing the observed degree of conservation in the
alignment as a whole, as discussed in Nicholas et al. (2002).