Page 69 - Annual report 2021-22

Annual Report 2021-22 |

discovery. A unified BC resource with ease of navigation, analysis, and visualization of the results, all
at one place will help researchers to expedite the analyses for better insights. But data acquisition,
harmonization, storage, and visualization can be challenging and require computational automation
with domain expertise. For instance, fetching bulk and single-cell datasets from Gene Expression
Omnibus (GEO) resource, harmonizing, visualization along with metadata require advanced data
analytical approaches. They are developing semi-automated pipelines to merge gene expression 52
series matrices with available metadata to be stored in the backend structured query language-based
system. The respective gene count matrix will be fed into the database which will be constructed and
managed using supporting modalities viz. PHP, Apache, SQL, and R Shiny applications. Visualization is
planned in the form of plots and graphs generated as a result of applied statistical tests with
significance level. Based on the data collated in this resource, they will be able to perform supervised,
semi-supervised and unsupervised machine learning approaches for the identification and in silico
validation of diagnostic, predictive and prognostic biomarkers, wherever sufficient data points will be
available. Furthermore, they are working on identification of the co-expression based conserved
modules in independent datasets with special focus on immune infiltration and tumor
microenvironment to discover immune hot and cold regions in patient-specific manner to aid in
precision therapy. In summary, the group is working on development of the breast cancer atlas where
user can search, browse, and analyze the reported/novel potential multi-omics biomarkers. This
resource will comprise of bulk and single-cell datasets encompassing genomics, transcriptomics, and
proteomics data of the samples which will not only be helpful in clinical settings for identifying patient-
specific biomarker using machine/deep learning algorithms but also in generating the data-driven
hypotheses for patient stratification, drug responsiveness, and in estimating survival groups.

“Light brings us the news of the universe.” – William Bragg