High-throughput scientific discoveries

High-throughput experiments have flooded biomedical research community with large volumes of data, causing the so called big data challenge. The data are not knowledge if not being analyzed, and turning the big data to scientific discoveries has become one of the bottlenecks of biomedical research. The main reason for the existence of the bottleneck is that we are doing low-throughput discoveries while data are being generated in a high-throughput fashion. As a result, there are more and more under-analyzed data. On our platform, high-throughput scientific discoveries can be made using integrated data and analysis tools. We are making scientific discoveries by ourselves for the following reasons:

  1. We illustrate the effectiveness of our platform by making scientific discoveries using the integrated data and tools on the platform. Each study will serve as a showcase of how one can use our platform to make high-quality, publishable discoveries.
  2. By being real users of our own system, we can keep improving and refining it, to meet the ever changing customer needs.
  3. The scientific discoveries made by us may serve as good starting points for other researchers to perform more indepth studies or directly commercialize them. If you are interested in collaborating with us to publish the results or perform experimental validations, please contact us.

Below we list some of the discoveries we have made.

Integrative differential gene expression analysis

Differential gene expression analysis compares gene expression between two (or among multiple) groups of subjects. It is one of the most common tasks in genomics research. We perform gene level, gene set level, and pathway/network level differential expression analysis and put discoveries in the context of existing knowledge to help our customers obtain deep understanding of the significance of the results. We have applied such analysis in the study of cancer health disparity using TCGA data.

For example in BRCA
African American vs. Caucasian American
Asian American vs. Caucasian American

Integrative differential DNA methylation analysis

Differential DNA methylation analysis is the most common task in analyzing DNA methylation data. We perform analysis to identify both differential methylation sites and regions and map them to genes. The genes are then studies through gene set, pathway, and network based methods. We have applies such analysis in the study of cancer health disparities using TCGA data.

For example in BRCA
African American vs. Caucasian American
Asian American vs. Caucasian American

Cancer Heterogeneity using Biclustering

A better understanding of cancer heterogeneity is the key for developing effective precision medicine for cancer treatment. We have developed a novel sequential biclustering method and applied the method to study the heterogeneity of cancer immune evasion mechanisms. We have identified several immune evasion subtypes for both breast cancer and prostate cancer using RNA-seq data at TCGA. These studies allow rational design of combination immunotherapies and clinical trials.

For example
SBic

Predictive models

Building predictive models for important clinical variables is another commonly encountered problem in translational biomedical science, such as prediction of survival, treatment response, etc. We have developed a comprehensive tool for building machine learning models using genomic data.