Identifying Protein Biomarkers for Ovarian
Cancer
Research
Abstract
Ovarian
cancer is the leading cause of gynecologic cancer deaths among American
women according to the American Cancer Society. The reason is attributed to
the fact that most cases get diagnosed at late stages of cancer at which
point chances of successful treatment are very low. Currently, there is no
effective non-invasive method to screen for ovarian cancer. Surgery is the
only reliable way to distinguish between benign and malignant disease,
resulting in several unnecessary surgeries. The objective of this research
is to develop an optimal biomarker prediction model to diagnose ovarian
cancer. The model will help develop strategies for detection that can be
both highly accurate and minimally invasive. Protein Fractionation- 2
Dimension (PF2D) is a system of hardware, software and chemistry that can
be used to identify protein levels from both urine and blood samples.
However, there are major challenges in analyzing the data resulting from
PF2D. Some of these challenges are removal of noise, baseline correction,
spectra normalization, peak extraction, protein biomarkers identification
and finally pattern generation for classification. For this research, PF2D
data from 10 urine samples of cases paired with their corresponding
controls are being analyzed. We address the above mentioned challenges by
developing optimization and data mining models that can efficiently
classify cases and controls.
Publications
Gopalappa, C., Das T.K., Thomas,
E., Koomen, J. and Sutphen,
R. 2011. Analytical identification
of small protein peaks. In review with Bioinformatics.
(For Codes and data sets click here)
Conference
Presentations
Gopalappa, C. and Das, T. K. Analytical identification of small protein peaks. INFORMS Annual Meeting 2010, Austin, TX.
Gopalappa, C. and Das, T. K. Optimization Models to Identify
Protein Biomarkers for Ovarian Cancer Diagnosis. INFORMS Annual Meeting
2009, San Diego, CA.
|
