Dr. Alioune NgomSchool of Computer Science, University of Windsor
Detecting driver genes and network biomarkers of breast cancer subtypes
Co-Investigators and Collaborators:
- Dr. Luis Rueda, School of Computer Science, University of Windsor
- Dr. Caroline Hamm, Windsor Regional Hospital/Windsor Cancer Program, Department of Biological Sciences, University of Windsor
- Dr. Lisa Porter, Department of Biological Sciences, University of Windsor
- Dr. Iman Rezaeian, School of Computer Science, University of Windsor (November, 2014)l
EVIDENCE OF PROGRESS
We have devised computational approaches for identifying the differentially expressed network of biomarkers of (NBs) of the breast cancer (BC) subtypes as well as for detecting the drivers of BC subtypes. This is achieved by:
(1) Appropriately integrating information contained within (a) primary transcriptomic data (gene expression data) and genomic data (copy-number variation, copy-number aberration, and single-nucleotide polymorphism data) with (b) a secondary interactomic data (protein-protein interaction network) which provide additional knowledge on the relationships between the genes contained in both the genomic and transcriptomic data.
(2) Using new machine learning algorithms developed in our laboratory for gene selection, for network biomarker identification, and for classification which takes as input both (a) the expressions of genes contained in the network biomarker and (b) the tumor samples of given patients to predict their correct subtypes.
Our best computational model has resulted in a set of small diagnostic network biomarkers capable of predicting the correct subtypes of a given but arbitrary breast cancer patients, with predictive accuracy ranging from 92% to 99%. Starting from this new model, we have also devised additional computational techniques to detect the driver genes of each breast cancer subtype given their identified network biomarkers. We have detected thus far a list of about 100 subtype drivers and they are currently being validated for their biological significance and soundness.
Our goal for the third year is to: (1) finish obtaining all the drivers and then validate them via functional analysis of cellular pathways and gene or network ontologies; (2) validate the identified network biomarkers statistically for their biological soundness and then interrogating their biological functions through functional enrichment techniques, ontology methods, and analysis of their annotations from cancer databases; and (3) biologically validate the driver functions of the identified network biomarkers by both Dr Lisa Porter and (possibly) Dr Caroline Hamm.
MEASURES OF PROGRESS
A) Manuscripts Published, Submitted, In Preparation (1): One manuscript is in preparation for submission in 2017
B) Conference Presentations/Proceedings (4): Four presentations were given during 2015 and 2016 in Minsk, Belarus, Niagara Falls, Canada, Washington DC, USA, and Buenos Aires, Argentina
C) Conference Poster Presentations (6): Six poster presentations were given made during 2014, 2015 and 2016: three in Windsor, Canada, two in Pittsburg, USA, and one in Toronto Canada
D) Grants Received Based on Data Obtained from S4H Research (1):
• Integrative Network-Based Machine Learning Approaches for Cancer Bioinformatics and Bio-Molecular Network Reconstruction (April 11, 2016 — RGPIN-2016-05017). Natural Sciences and Engineering Research Council of Canada (NSERC), Discovery Grant Program. $90,000CAD ($18,000CAD/year). Principal Investigator: Alioune Ngom.
E) Training of Highly Qualified Personnel:
Post-Doctoral Researcher (1): Dr. Iman Rezaeian (Fall 2014)
Doctoral Students (3): Forough Firoozbakht (Fall 2014); Huy Quang Pham (Fall 2015): Hetal Rajpura (Fall 2016).
Graduate Students (1): Sheikh Jubair (Fall 2016).
Senior Undergraduate Students (3): Michele D’Agnillo (Biological Sciences student (Winter 2015); Maher Hussain and David Valleau (Computer Science students (Fall 2016).
F) Other Measures: New Collaborative Endeavors
1. Dr. Peter Rogan (School of Medicine, University of Western Ontario). Our joint-research (together with Dr Luis Rueda) involves predicting response to chemotherapies in breast cancer by machine learning.
2. Dr. Ram Samudrala (School of Medicine, State University of New York in Buffalo [SUNY-Buffalo]). Our joint-research involves developing a platform that will screen and rank every existing cancer drug or compound (with establish side effects profile) for every cancer disease or indication, using machine learning tools as well as structural bioinformatics tools.