say you’re watching the Super Bowl and
you see an ad for a prescription drug
we’ve all seen them the beginning paints
an optimistic picture of the drugs
incredible curing capabilities followed
by a lengthy list of side effects
ranging from dizziness to death when
someone takes a prescription drug
they’re also taking a potentially fatal
gamble with over 70% of Americans on at
least one prescription variable
responses to medication are a huge
problem adverse drug events are
responsible for approximately 50% of all
annual Hospital all annual
hospitalizations and are the fourth
leading cause of death in the u.s.
adverse drug events are responsible for
approximately 1.5 million annual
hospitalizations that’s 3 percent of all
annual hospitalizations meaning that
every minute three people are
hospitalized with an adverse drug
reaction why might a drug cure one
person but poison another I decided to
pursue a project in bioinformatics the
science of analyzing genetic data to
explore this question i hypothesized
that a person’s response to a drug is
influenced by their genome or all their
DNA to develop my idea
I thought about DNA and how drugs work
what’s a genetics have to do with a
drugs response once ingested the
digestive system breaks down a drug into
molecules that flow through the
bloodstream a person responds to a drug
when drug molecules bind to body
molecules and trigger a certain response
this response procedure is known as a
self signaling pathway cell signaling
pathways are like locking key mechanisms
where the part of the drug molecule is
the key that fits into part of the body
molecule when molecules bind together
they unlock the door to a physical
response to a drug by sending a signal
to the body chemicals in the body called
enzymes speed up the reactions between
drug molecules and body molecules and
can therefore affect the cell signaling
pathway DNA contains the instructions
for the body to produce enzymes enzymes
are made of proteins and your unique set
of proteins is what makes you you
mutations or changes in the DNA code
could change what proteins and enzymes
are produced therefore a change in the
enzyme could change a signal that the
drug sends to the body so enzymes maybe
it was in mutations may be responsible
for side effects when I read about
genome sequencing I thought wow what if
there’s a way to analyze a genome to
find a mutation then there might be a
way to correlate these mutations to side
effects I researched mutations in
enzymes critical to the absorption of
drugs lipitor plavix and tegretol I
focused on these drugs because they have
been well studied using public databases
I found mutations linked to drug in
efficacy and side effects such as
excessive bleeding for lipitor plavix
however knowing the mutation only solves
half the problem DNA code is composed of
building blocks known as basis these
bases are like the alphabet of the DNA
code for proteins to know if there is an
abnormal mutation in this DNA code
doctors need to search the whole genome
that’s six billion bases long when I saw
this challenge I added to my hypothesis
and predicted that I could innovate a
computer algorithm or a problem solving
procedure to find a mutation this
sparked my next idea I could innovate a
string searching algorithm string
searching algorithms are computer
programs that find a string of letters
in a text the genome is the text and the
mutation is the string how will string
searching algorithms save lives these
algorithms can find mutations allowing
doctors to search a database for
mutations compare them with ones known
to cause side-effects and prescribe a
possibly life-saving drug my innovation
is a string searching algorithm that
finds mutations linked to drug responses
this can help solve the problem of
variable responses to prescription drugs
I create a design criteria for my
innovation it needs to be functional
feasible and fast why is there a need
for speed a fast algorithm could be
life-saving in emergency situations such
as seizure or allergic shock when the
correct drug must be prescribed
immediately to address this need I
tested several algorithms and narrowed
it down to to apostille eco Giancarlo
and reverse factor string searching
algorithms in general work to find a
pattern P within a text team apostille
eco Giancarlo uses an efficient
algorithm to store characters and find a
match instead of comparing every
character in pattern P with every
character in text T resulting in
time-consuming operations apostille eco
Giancarlo can skip mismatched alignments
when a match is not found reverse factor
is faster because it uses a similar
method to apostille eco Giancarlo
but stores data more efficiently so a
type of array that allows the computer
to quickly access data values from
memory however I knew that the existing
algorithms couldn’t find a mutation fast
enough so I innovated parallel versions
of them a parallel program can run on
multiple computer processors how does
parallelism make an algorithm faster a
parallel program can run on multiple
computer processors parallel computing
is like weaving a rug where the rug is
the task or searching the genome and the
workers are the processors if one worker
does the job it could take that worker a
month but if ten workers do the job the
rug would only take a few days to finish
in a nutshell parallelism makes
processors work together reducing the
runtime I found that paralyzing the
algorithms with four processors
accelerated each sequential one by 400%
parallel reverse factor was double the
speed of parallel apostille eco
Giancarlo accelerating the original
algorithm by 8
percent though I obtained my results in
milliseconds with an 800,000 based
excerpt from chromosome 1 each algorithm
would take hours to run using a whole
genome the human genome is 6 billion
7,500 times longer to run than in my
experiments
doctors would run an algorithm to search
for multiple mutations in the whole
genome which could take several hours
this might be too late for patients in
emergency situations therefore an
accelerated algorithm like mine will
help save lives by allowing doctors to
quickly predict a patient’s response to
a drug I became a finalist in the
discovery 3m young scientist challenge
for this research I was able to develop
my innovation through a summer
mentorship with 3m scientist John
Henderson industry leader and medication
risk management software company tabula
rasa healthcare invited me to present at
their headquarters though my algorithm
was effective I thought I could improve
the feasibility since the procedure for
hospitals to access genomes is still not
firmly established now the good news is
genome sequencing is becoming more
widespread and was expensive to obtain
some hospitals are already sequencing
every newborns genomes demonstrating the
increasing feasibility of personalized
medicine however accessing genomes and
translating complex medical tests into a
I created a prototype of a graphical
user interface that shows how I envision
using my innovation to help doctors
access genomes
it shows how I envision using my
parallel reverse factor algorithm and
medication risk management software to
help save lives now I’m going to show
you my prototype my proposed prototype
would allow doctors to access genomes
find mutations and then link them to
drug responses imagine a patient
collapses with a seizure the patient
needs a drug right away the doctor
enters the patient’s diagnosis candidate
drugs and sequence genome the doctor
then runs my parallel reverse factor
algorithm to gain information about a
potential prescription in the time of
this eight-minute talk twenty four
people were hospitalized with an adverse
drug reaction my innovation has the
potential to reduce these
hospitalizations and improve health by
advancing the field of personalized
medicine I envision a future where
treatment is tailored to one’s
individual genome where someone can take
a drug and not just hope for but expect
a positive outcome bioinformatics
research similar to mine can soon
transform this vision into reality by
allowing doctors to make the pill fit
the ill thank you [Applause]