Graduate / A step closer to treating cancer - PHD SOP for Computational Data Enabled Science and Engineering [4]

For many years now, Scientists have struggled to help medical practitioners treat their patients according to their symptoms and provided customized healthcare on a personal basis. However, how personal can medicine get? In 2003, researchers obtained a complete human genome from which the sequence and map of all genes in the human body can be used as a reference. With this development we are a step closer to treating cancer and other diseases.

Most cancer treatments engage in trial and error basis of treatment which puts so much of pain, and risk on the patient. I worked on a similar problem in my role as a research assistant in a project funded by the NIH ******* Clinical Translational Research Center. It was on the use of metabonomics for the early detection of ovarian cancer. We worked on the development of various multi-class modelling techniques like support vector machines, random forests, k nearest neighbours and partial least squares using RNA-seq data from XYZInstitute that helped estimate accurate sensitivity, specificity, and positive predictive values of the tumour and develop predictive models for cancer.

I have worked on DNA methylation, RNA expression, clinical patient data based on the Pancreatic Adenocarcinoma dataset through the cancer genome atlas (TCGA). The study which was a two-step analysis included a pilot analysis that used supervised statistical analysis to identify gene features or RNA transcripts which are differentially expressed between covariates of particular interest yielding statistical significance. A follow up analysis was performed, which yielded us significant genes and proved to be an important integration analysis of multiple cancer and tissue types, an important area for bioinformatics.

In cancer, every tumour its own unique genetic makeup. In cancer, every tumour its own unique genetic makeup. Genomic data is important to comprehensively treat a patient. Genome sequences can be put in cloud systems like Hadoop clusters unlike traditional Relational databases where there is a restriction on the input format. I like to work on data to make it timely, fresh and of high quality to be available at the right format and at the right place like a good course of treatment and not just focus on optimizing traditionally managed data.

In my graduate studies, under the Bioinformatics and Biostatistics program, I took Data Mining subjects that dealt with supervised and unsupervised methods of learning that helped shape my interest in Data Science. I worked on a dataset that analyzed the malignant and benign prostate tissues that provided insight into a gene expression signature for prostate cancer. The goal was to conduct several prediction algorithms including principal component regression, elastic nets, partial least squares on the samples containing gene expression profiling by array to accurately predict prostate cancer. It was based on variables like factors that included disease state, and genotype/phenotype variation.

I currently work on building probabilistic graphic models by constructing metabolic models of genes. It involves understanding the metabolic pathways and testing hypothesis using high throughput data from human reconstruction models. Here the metabolic network models are used to investigate genes significant to Alzheimer's disease. A generic constraint based model of cancer metabolism is used to help predict important genes. We incorporate Bayesian statistics for belief propagation and test for statistical significance.

These day, large amounts of data is being generated from devices that track a person's health. I plan to research extensively using big data in these forms and use them to solve significant risk factors. I like to help predict the possible outcomes in an experiment/course of treatment of a disease and thus help contribute in increasing the survival statistics of a patient. I hope to work towards improving healthcare analyzing practices by recognizing the weakness and leverage the strengths of various algorithm approaches. I think the Computational and Data-Enabled Science and Engineering program will help me in pursuing this goal of mine.

It interests me to learn and excel in a place that has a huge intertwining of cultures and knowledge. I believe that it helps me discover my place in research of curing diseases. I am deeply influenced by the work of Professor ABC at the Department of Biostatistics and I plan to work under her for tackling big data problems in bioscience by employing mathematical modeling.