What Is Bioinformatics
What Is Bioinformatics
What Is Bioinformatics
B. J ayaram, N. Latha, Pooja Narang, Pankaj Sharma, Surojit Bose, Tarun J ain, Kumkum
Bhushan, Saher Afshan Shaikh, Poonam Singhal, Gandhimathi and Vidhu Pandey
Department of Chemistry &
Supercomputing Facility for Bioinformatics & Computational Biology,
Indian Institute of Technology, Hauz Khas, New Delhi - 110016, India.
Email: bjayaram@chemistry.iitd.ac.in
Web site: www.scfbio-iitd.org
I. What is Bioinformatics?
Bioinformatics is an emerging interdisciplinary area of Science & Technology
encompassing a systematic development and application of IT solutions to handle
biological information by addressing biological data collection and warehousing, data
mining, database searches, analyses and interpretation, modeling and product design.
Being an interface between modern biology and informatics it involves discovery,
development and implementation of computational algorithms and software tools that
facilitate an understanding of the biological processes with the goal to serve primarily
agriculture and healthcare sectors with several spin-offs. In a developing country like
India, bioinformatics has a key role to play in areas like agriculture where it can be used
for increasing the nutritional content, increasing the volume of the agricultural produce
and implanting disease resistance etc.. In the pharmaceutical sector, it can be used to
reduce the time and cost involved in drug discovery process particularly for third world
diseases, to custom design drugs and to develop personalized medicine (Fig. 1).
Gene finding Protein structure prediction Drug design
Figure 1 : Some major areas of research in bioinformatics and computational biology.
Figure 2 : The tree of life depicting evolutionary relationships among organisms from the
major biological kingdoms. A possible evolutionary path from a common ancestral cell to
the diverse species present in the modern world can be deduced from DNA sequence
analysis. The branches of the evolutionary tree show paths of descent. The length of paths
does not indicate the passage of time and the vertical axis shows only major categories of
organisms, not evolutionary age. Dotted lines indicate the supposed incorporation of
some cell types into others, transferring all of their genes and giving the tree some web-
like features. (Source : A Alberts, D Bray, J Lewis, M Raff, K Roberts & J D Watson, Molecular
Biology of the Cell, p38, Garland, New York (1994)).
It is assumed that life originated from a common ancestor and all the higher
organisms evolved form a common unicellular prokaryotic organism. Subsequent
division of different forms of life from this makes the diversity in the morphological and
genetic characters (Fig. 2).
Figure 3 : (A) An animal cell. The figure represents a rat liver cell, a typical higher
animal cell in which features of animal cells are evident such as nucleus, nucleolus,
mitochondria, Golgi bodies, lysosomes and endoplasmic reticulum (ER). (Source:
www.probes.com/handbook/ figures/0908.html) (B). A plant cell (cell in the leaf of a higher
plant). Plant cells in addition to plasma membrane have another layer called cell wall,
which is made up of cellulose and other polymers where as animal cells have plasma
membrane only. The cell wall, membrane, nucleus chloroplasts, mitochondria, vacuole,
ER and other organelles that make up a plant cell are featured in the figure (Source:
http://www.sparknotes.com/biology/cellstructure/celldifferences/section1.html).
(A). Animal cell (B). Plant cell
The common basis to all these diverse organisms is the basic unit known as the
cell (Fig. 3). All cells whether they belong to a simple unicellular organism or a complex
multicellular organism (human adults comprise ~30 trillion cells), possess a nucleus
which carries the genetic material consisting of polymeric chains of DNA (deoxyribo
nucleic acid), holding the hereditary information and controlling the functioning. Several
challenges lie ahead in deciphering how DNA, the genetic material in these cells
eventually leads to the formation of organisms (Fig. 4).
CELL
TISSUE
ORGAN
ORGANISM
Figure 4 : Levels of organization. The entire DNA content in a cell is called the genome.
In spite of the complex organization, cells of all organisms possess different
The entire protein content in a cell is called the proteome. Cellome is the entire
complement of molecules, including genome and proteome within a cell. Tissues are
made of collections of cells. Tissue collections make organs. An organism is a collection
of several organ systems.
molecules of life for the maintenance of living state. Theses molecules include nucleic
acids, proteins, carbohydrates and lipids (Fig. 5).
Nucleic Acids
Carbohydrates
Proteins
Lipids
Figure 5 : Biom
All organisms self re genetic material DNA, the
olecules of life
plicate due to the presence of
polynucleotide consisting of four bases Adenine (A), Thymine (T), Guanine (G) and
Cytosine (C) (Fig. 6). The entire DNA content of the cell is what is known as the genome.
The segment of genome that is transcribed into RNA is called gene. So we can say that
hereditary information is transferred in the form of genes contained on the four bases.
Understanding these genes is one of the modern day challenges. Why only five percent of
the entire DNA is in the form of genes and what is the rest of the DNA responsible for,
under what conditions genes are expressed, where, when and how to regulate gene
expression are some unsolved puzzles.
Figure 6 : DNA and its alphabets A, G, C & T the nucleic acid bases. (Source: Molecular
rmation present in DNA is expressed via RNA molecules into proteins
biology of the cell, Garland Publishing. Inc. New York, 1994 )
The info
which are responsible for carrying out various activities. This information flow is called
the central dogma of molecular biology (Fig. 7). Potential drugs can bind to DNA, RNA
or proteins to suppress or enhance the action at any stage in the pathway.
Figure 7 : Central dogma of modern biology
enome Analysis
f genome called genes determine the sequence of amino acids in
otein
n order to identify all
average of just 2% i.e. just about 160 enzymes.
G
Segments o
pr s. The mechanism is simple for the prokaryotic cell where all the genes are
converted into the corresponding mRNA (messenger ribonucleic acid) and then into
proteins. The process is more complex for eukaryotic cells where rather than full DNA
sequence, some parts of genes called exons are expressed in the form of mRNA
interrupted at places by random DNA sequences called introns. Of the several questions
posed here, one is that how some parts of the genome are expressed as proteins and yet
other parts (introns as well as intergenic regions) are not expressed.
Scores of genome projects are being carried out world wide i
the genes and ascertain their functions in a specified organism. Human genome project
is one such global effort to identify all the alphabets on the human genome, initiated in
1990 by the US government. A comparison of the genome sizes of different organisms
(Table 1) raises questions like what types of genetic modifications are responsible for the
four times large genome size of wheat plant and seven times small size of the rice plant
as compared to that of humans. Mice and humans contain roughly the same number of
genes about 28K protein coding regions. The chimp and human genomes vary by an
RNA
Genome
Gene = DNA
DNA binding drugs
Protein
Gene therapy
Primary Sequence
RPRTAFSSEQLARLK
REFNENRYLTERRR
QQLSSELGLNEAQIK
IWFQNKRAKIKKS
RNA binding drugs Drugs
Inhibitors/activators
Table 1 : Genome sizes of some organisms
Organism Genome size (Mb)
(Mb=Mega base)
Eschericia coli 4.64
M t s uberculosi 4.4
H.Influenza 1.83
Homo sapiens (humans) 3300
Mouse 3000
Rice 430
Wheat 13500
(source : http://users m/jkimball.ma.ultranet/BiologyPages nomeSizes.html)
Table 2 : Specific genetic disorders
.rcn.co /G/Ge
Genetic Disorder Reason
Huntingtons Disease Excessive repeats of a three-base sequence, "CAG" on
chromosome 4.
P arkinsons Disease Variations in genes on chromosomes 4,6.
Sickle Cell Disease Mutation in hemoglobin-b gene on chromosome 11
Tay-Sachs Disease Co 15 ntrolled by a pair of genes on chromosome
Cystic Fibrosis Mutations in a single (CFTR) gene
Breast Cancer M utation on genes found on chromosomes 13 & 17
Leukemia Exchange arms of of genetic material between the long
chromosome 6 & 22.
Colon cancer P roteins MSH2, MSH6 on chromosome 2 & MLH1 on
chromosome 3 are mutated.
Asthma Disfunctioning of genes on chromosome 5,6,11,14&12.
Rett Syndrome Disfunctioning of a gene on the X chromosome.
Bruk oma itt lymph Translocations on chromosome 8
A lzheimer disease Mutat 9 & ions on four genes located on chromosome 1, 14, 1
21.
Werner Syndrome Mutat me 8. ions on genes located on chromoso
A ngelman Syndrome Deletion of a segment on maternally derived chromosome 15.
Best disease Mutat e 11. ion in one copy of a gene located on chromosom
Pendred Syndrome Defective gene on chromosome 7.
Dias sia trophic dyspla Mutation in a gene on chromosome 5
(Source:ht h.gov/)
tp://www.ncbi.nlm.ni
Several genetic disorders like Huntingtons disease, Parkinsons disease, sickle
cell anemia etc. are caused due to mutations in the genes or a set of genes inherited from
one generation to another (Table 2). There is a need to understand the cause for such
disorders. Why nature carries such disorders and how to prevent these are some of the
areas where extensive research endeavor has to be invested.
An understanding of the genome organization can lead to concomitant progresses
in drug-target identification. Comparative genomics has become a very important
emerging branch with tremendous scope, for the above mentioned reasons. If the genome
for humans and a pathogen, a virus causing harm is identified, comparative genomics can
predict possible drug-targets for the invader without causing side effects to humans (Fig.
8).
Potential Areas for Drug Targets = H
c
I
H = Human Genome / Proteome (Healthy Individual)
I = Genome / Proteome of the Invader / Pathogen
Figure 8 : Comparative genomics for drug target identification an application of
Bioinformatics
Over the past two decades genetic modification has enabled plant breeders to
develop new varieties of crops like cereals, soya, maize at a faster rate. Some of these
called as transgenic varieties have been engineered to possess special characteristics that
make them better. Recently efforts are on in the area of utilizing GM (genetically
modified) crops to produce therapeutic plants.
Another area where bioinformatics techniques can play an important role is in
SNP (Single nucleotide polymorphisms) discovery and analysis. SNPs are common DNA
sequence variations that occur when a single nucleotide in the genome sequence is
changed. SNPs occur every 100 to 300 bases along the human genome. The SNP
variants promise to significantly advance our ability to understand and treat human
diseases.
Given the whole genome of an organism, finding the genes (gene annotation) is a
challenging task. Various approaches have been utilized by different groups for this.
These approaches are typically database driven which rely on the already known
forma
r which attempts to capture forces responsible for DNA structure and protein-DNA
interactions walks along the length of the genome distinguishing genes from non-genes.
As of now, the physico-chemical model (ChemGene 1.0) is able to distinguish
successfully genes from non-genes in 120 prokaryotic genomes with fairly good
specificities and sensitivities.
Protein Folding
Proteins are polymers of amino acids with unlimited potential variety of structures
and metabolic activiti e dimensional shape
called its confo its shape and
genetic code. Substitution of a single amino acid can cause a major
tist made some significant contributions
energy which is about 10 to 15 kcal/mol relative to the unfolded state. How does a given
sequence of amino acids fold into a specific conformation as soon as it is conceived on
in tion. The Markov model based methods utilize short range correlations between
bases along the genome. Other methods based on Fourier transform techniques
emphasize global correlations.
At IIT Delhi, we are attempting to develop a hypothesis driven physico-chemical
model for genome analysis and for distinguishing genes from non-genes. In this model, a
vecto
es. Each protein possesses a characteristic thre
rmation. Sequence of amino acids of protein determines
function. This in turn is genetically controlled by the sequences of bases in DNA of the
cell through the
alteration in function. Study of homologous proteins from different species is of interest
in constructing taxonomic relationships. Proteins may be classified into structural
proteins, enzymes, hormones, transport proteins, protective proteins, contractile proteins,
storage proteins, toxins etc..
Prof G N Ramachandran, an Indian scien
towards an understanding of the secondary structure of proteins. His fundamental work in
this area is remembered in the form of Ramachandran maps.
Protein folding can be considered to involve changes in the polypeptide chain
conformation to attain a stable conformation corresponding to the global minimum in free
the ribosomal machinery using the information on mRNA in millisecond timescales, is
the problem pending a resolution for over five decades. For a 200 amino acid protein with
just two conformations per amino acid, a systematic search for this minimum among all
possible 2
200
conformations, will take approximately 3x10
54
years which is much longer
than the present age of the universe. Despite this innumerable number of conformations
to search, nature does it in milliseconds to seconds (Fig. 9).
Figure 9 : The protein folding problem (Source: press2.nci.nih.gov/.../ snps_cancer21.htm)
Solution to the protein folding problem has an immense immediate impact on
society. In biotech industry, this can be helpful in the design of nanobiomachines and
sed drug discovery wherein
the stru
biocatalysts to carry out the required function. Pulp, paper and textile industry, food and
beverages industry, leather and detergent industry are among the several potential
beneficiaries. Other important implications are in structure ba
cture of the drug target is vital (Fig. 10). Structures of receptors a major of class
of drug targets - are refractory to experimental techniques thus leaving the fold open to
computer modeling.
Figure 10 : Drug targets (Source: Drews J ., Drug Discovery: A historical perspective, Science, 2000,
287: 1960-1964)
There is an urgent demand for faster and better algorithms for protein structure
prediction. There are three major ways in which protein structure prediction attempts are
currently progressing viz. homology modeling, fold recognition and ab initio approaches.
Whereas the first two approaches are database dependent relying on known structures and
folds of proteins, the third one is independent of the databases and starts from the
physical principles. Although the ab initio techniques till date are able to predict only
small proteins, because of their first principle approach, they have the potential to predict
new / novel folds and structures. On account of the complexity of the problem,
computational requirement for such a prediction is a major issue which can easily be
appreciated by the estimates of time required to fold a 200 amino acid protein which
evolves ~10
-11
sec per day per processor according to Newtons laws of motion. This will
require approximately a million years to fold a single protein. If one can envision a
million proc tein can be folded in one ear computer
e. In this Blue Gene
essors working together, a single pro y
tim spirit, the IBM has launched a five year project in the year 1999
to address the complex biomolecular phenomena such as protein folding.
At IIT Delhi, we are developing a computational protocol for modeling and
predicting protein structures of small globular proteins. Here a combination of
bioinformatics tools, physicochemical properties of proteins and ab initio approaches are
used. Starting with the sequence of amino acids, for ten small alpha helical proteins,
structures to within 3-5 of the native are predicted within a day on a 50 ultra sparc III
Cu 900 MHz processor cluster. Attempts are on to further bring the structures to within
<3 of the native structures via molecular dynamics and Metropolis Monte Carlo
simulations.
Drug Design
As structures of more and more protein targets become available through
crystallography, NMR and bioinformatics methods, there is an increasing demand for
computational tools that can identify and analyze active sites and suggest potential drug
molecules that can bind to these sites specifically (Fig. 11).
Figure 11 : Active-site directed drug-design
Also to combat life-threatening diseases such as AIDS, Tuberculosis, Malaria etc.,
a global push is essential (Table 3). Millions for Viagra and pennies for the diseases of
the poor is the current situation of investment in Pharma R&D.
Active Site
Table 3 : WHO Calls for Global Push Against AIDS & Tuberculosis & Malaria
(Source : www.globalhealth.org, www.thenation.com )
Time and cost required for designing a new drug are immense and at an
unacceptable level. According to some estimates it costs about $880 million and 14 years
of research to develop a new drug before it is introduced in the m
Lead Generation & Lead Optimization
3.0 yrs 15%
Preclinical Development
1.0 yrs 10%
Phase I, II & III Clinical Trials
6.0 yrs 68%
FDA Review & Approval
1.5 yrs 3%
Drug to the Market
14 yrs $880million
arket (Table 4).
Table 4 : Cost and time involved in drug discovery
Target Discovery
2.5 yrs 4%
(Source : PAREXEL, PAREXELs Pharmaceutical R&D Stastical Sourcebook, 2001, p96)
illion die each year due to diseases like diarrhoea, small pox, polio,
river blindness, leprosy.
cost $ 12 billion to fight the disease of poverty
(AIDS medication about $15K per annum)
Millions for Viagra, Pennies for the Diseases of the Poor
Of all new medications brought to the market (1223) by Multinationals from
1975 only 1% (13) are for tropical diseases plaguing the third world.
Life style drugs dominate Pharma R&D
(1) Toe nail Fungus (2) Obesity
(3) Baldness (4) Face Wrinkle
(5) Erectile Dysfunction (6) Separation anxiety of dogs etc.
Nearly 6 m
Estimated
A new economic analysis
Healthy people can get themselves out of poverty.
Relieving a population of burden of the diseases for 15 to 20 years will give a
huge boost to economic development.
Nearly 6 million die each year due to diseases like diarrhoea, small pox, polio,
river blindness, leprosy.
cost $ 12 billion to fight the disease of poverty
(AIDS medication about $15K per annum)
Millions for Viagra, Pennies for the Diseases of the Poor
Of all new medications brought to the market (1223) by Multinationals from
1975 only 1% (13) are for tropical diseases plaguing the third world.
Life style drugs dominate Pharma R&D
(1) Toe nail Fungus (2) Obesity
(3) Baldness (4) Face Wrinkle
(5) Erectile Dysfunction (6) Separation anxiety of dogs etc.
Estimated
A new economic analysis
Healthy people can get themselves out of poverty.
Relieving a population of burden of the diseases for 15 to 20 years will give a
huge boost to economic development.
Intervention of computers at some plausible steps is imperative to bring down the cost
an
Fig re 12
E
for a given drug
d time required in the drug discovery process (Fig. 12, Table 5).
u : Potential areas for In silico intervention in drug-discovery process
Table 5 : High End Computing Needs for In Silico Drug Design
stimates of current computational requirements to complete a binding affinity calculation
(Source : http://cbcg.lbl.gov)
In silico methods can help in identifying drug targets via bioinformatics tools.
They can also b g / active sites,
generate candidate m
Modeling complexity Method Size of Required computing
library time
Molecular Mechanics SPECITOPE 140,000 ~1 hour
Rigid ligand/target LUDI 30,000 1-4 hours
CLIX 30,000 33 hours
Molecular Mechanics Hammerhead 80,000 3-4 days
Partially flexible ligand DOCK 17,000 3-4 days
Rigid target DOCK 53,000 14 days
e used to analyze the target structures for possible bindin
olecules, check for their drug-likeness, dock these molecules with
Molecular Mechanics
Fully flexible ligand
Rigid target
ICM 100,000 ~1 year
(extrapolated)
Molecular Mechanics
Free energy
perturbation
AMBER
CHARMM
1 ~several days
QM Active site and
MM protein
Gaussian,
Q-Chem
1 >several weeks
the target, rank them according to their binding affinities, further optimize the molecules
to improve binding characteristics etc..
-50
0
-30
-20
-10
0
10
20
30
B
i
n
d
i
n
g
F
r
e
e
E
n
e
r
g
y
(
k
c
a
l
/
m
o
l
)
Pursuing the dream that once the gene target is identified and validated drug
discovery protocols could be automated using Bioinformatics & Computational Biology
tools, at IIT Delhi we have developed a computational protocol for active site directed
drug design. The suite of programs (christened Sanjeevini) has the potential to evaluate
and /or generate lead-like molecules for any biological target. The various modules of
this suite are designed to ensure reliability and generality. The software is currently being
optimized on Linux and Sun clusters for faster and better results. Making a drug is more
like designing an adaptable key for a dynamic lock. The Sanjeevini methodology consists
of design of a library of templates, generation of candidate inhibitors, screening
candidates via drug-like filters, parameter derivation via quantum mechanical
calculations for energy evaluations, Monte Carlo docking and binding affinity estimates
e protocols
tested on Cyclooxygenase-2 as a target could successfully distinguish NSAIDs
(Nonsteroid on other
(red).
based on post facto analyses of all atom molecular dynamics trajectories. Th
al Anti-inflammatory drugs) from non-drugs (Fig 13). Validation
targets is in progress.
-4
Figure 13 : The active site directed lead design protocols developed from first principles
and implemented on a supercomputer could segregate NSAIDS (blue) from Non-drugs
Bioinformatics and Biodiversity
Biodiversity informatics harnesses the power of computational and information
technologies to organize and analyze data on plants and animals at the macro and at
genome levels. India ranks among the top twelve nations of the world in terms of
biological diversity (Fig. 14 and Fig15).
Himalayas - This majestic range of
mountains is the home of a diverse range
of flora and fauna. Eastern Himalayas is one
iversity hotspots in India.
Chilika - This wetland area is protected
under the Ramsar convention.
of the two biod
Sunderbans - The largest mangrove forest in Western Ghats - One of the two
India. biodiversity hotspots in India.
Thar desert ntrast to the Himalayan region.
(Source: http://edugreen.teri.res.in/explore/maps/biodivin.htm)
- The climate and vegetation in this area is a co
Figure 14 : Biodiversity in India
Figure 15: Biodiversity bioinformatics is essential to preserve the natural balance of flora
and fauna on the planet and to prevent extinction of species (Source: www.hku.hk/
ecology/envsci.htm).
Bioinformatics endeavors in India
Owing to the well acknowledged IT skills and a spate of upcoming software,
biotech and pharma industries and active support from Government organizations, the
field of Bioinformatics appears promising. The projections of the growth potential in
India in a global scenario however belie these expectations (Fig. 16).
Figure 16 : Growth potential for Bioinformatics based business opportunities in India
ccording to IDC (International Data Corporation, India.)
a
0
0.2
0.8
1
1.2
1.4
1.6
1.8
1
0.4
0.6
2
Year
2001 2005
Global
India
Creating the necessary infrastructure around specialists to meet the high
omputational power required for these activities (Table 6, Fig. 17) is one step in the
ght direction.
able 6 : Geno ate of computational
quirements
. Gene Prediction
c
ri
T me to drug discovery research: A rough estim
re
1
Homology/string comparison. 300 Giga flop
~3*10
9
bp
2. Protein Structure Prediction
- Threading 100 Giga flop
- Statistical Models
- Filters to reduce plausible structures
Molecular Dynamics
s per simulation 25000
Active site directed drug design
100 structures 30 Peta flop
1-ns simulation for structure refinement
Total Compute Time 5000ns
Number of atom
3.
Scan 1000 drug molecules/protein 18 Peta flop
Total Computational requirement to design one lead compound from genome
16
o design ten lead compounds per day (on a dedicated machine)
e requirement is 5.8 tera flops capacity.
ut of every 100 lead compounds, only one may become a drug, which further increases
e computer requirements)
3ns simulation per drug molecule
(Active site searches, docking, rate and affinity determinations etc.)
Total Compute Time 3000ns
25000 atoms per simulation
Summary
~50 Peta flop (5.x10 floating point operations)
T
th
(O
th
for bioinformatics
and scientific community from across the country,
l Private
iotechnology, Govt of India). It is envisioned that the
span all the 60+Bioinformatics Centres of the Department of Biotechnology
e 18).
Figure 17 : Challenges
To enable access by the student
the Supercomputing Facility at IIT Delhi is connected to Biogrid-India (a Virtua
Network of the Department of B
Biogrid will
with GBPS bandwidth and at least 10 teraflops of compute capacity soon (Figur
Transcription
Model
Education
RNA splicing
model
Signal
transduction
pathways
Protein:DNA A /RN
protein:protein
Protein structure
prediction
Drug Design
Protein
Evolution
Gene ontologies
Speciation
Challenges for
Bioinformatics
Figure 18 : Biogrid-India
Pro-active policies and conducive environment to researchers, entrepreneurs and
Industry will go a long way in steering India towards leadership in the area of
Bioinformatics and Computational Biology.
Acknowledgements.
BJ is thankful to the Department of Biotechnology, the Department of Science &
Technology, Council of Scientific & Industrial Research, Indo-French Centre for the
Promotion of Advanced Research and IIT Delhi administration for support over the years.