How the chromatin is organized within the nucleus and how its 3D architecture influences gene regulation, cell fate decisions, and evolution are some of the significant questions in biomedical research. The technological advancements to assess the 3D structure of chromatin at a population or at a single-cell level dictate the depth of our understanding of how it is established, restructured and maintained. In my talk, I will discuss the technological advancements to study the 3D genome architecture, e.g., Hi-C, and the recent developments in generating, analysing and interpreting such data. Furthermore, I will delve into the field of cerebral cortex development and discuss how the tools to study 3D chromatin are helping us answer some critical questions about the mechanisms of brain development and disease.

Blood-based early diagnosis of cancers and personalized treatment prioritization are among the top clinical oncology grand challenges. In the past few years, we leveraged single cell and bulk transcriptomics of cancer cells, circulating tumor cells (CTCs), and tumor-educated platelets (TEPs) to model various clinically relevant outcomes, such as the existence of cancer and drug resistance. In this talk, we will share some practical approaches to making clinically relevant problems computationally tractable and an excerpt of the key results.

Investigating network behaviour from host-pathogen interactions is challenging. Here, I will give talk about the application of integrated deep-learning and network biology to construct an immune-related network of host-pathogen interactions. The twin challenges of limited samples and protein-protein interactions in the deep learning makes an impending task to achieve higher accurate results. In order to address the challenges, we have developed the DLNet algorithm which tried to integrate the graph theory with a forest feature graph extractor with expression data as an input so that feature graph generated can be learned in a supervised manner. This protocol includes the steps to pre-process the interaction pairs and expression profile of host model treated with pathogen/control, feed as input for DLNet algorithm to rank genes based on their contribution to data classification. The DLNet algorithm was found to be superior performance with the state-of-the-art machine learning algorithms; network-guided forest (NGF), gradient boosting machine (GBM), and random forest (RF) on all datasets under study. And by large, the protocol developed can be applied to any kind of host-pathogen interactions provided the number of samples should be > 20.

Graphs are an essential part of many social and scientific studies, such as scholar networks, social networks, chemical structures, biological interactions, etc. However, performing analytics on ubiquitous graph data is non-trivial because of the unique characteristics of the graphs. From 2006-2016 shallow networks dominated the field. Recently, substantial research efforts have been made to develop deep learning-based techniques to perform analytics on graphs. In this lecture, I will present a brief history of the shallow methods and seminal works in deep learning on graphs with a hands-on session.

Deep-learning has revolutionized the entire world and is being considered as the turning point of civilization and mankind, with its impact on almost every field. This lecture session will first discuss on about the fundamentals of machine learning and how Deep-Learning is different. What makes it “Deep” and how it has become capable to achieve “anything”, while taking the route of our own experience with it in implementing deep-learning in the regulatory genomics areas like discovery of RBP-RNA interaction, discovery of long non-coding RNAs, and miRNAs discovery. In doing so, various deep learning architectures will be discussed along with how they see the input and how they decompose them, something which demarcates deep-learners from general machine learning approaches.

In India, non-communicable diseases (NCDs) are responsible for almost 46.6% of the health burden, leading to more than 10 million deaths and over 500 million DALYs. Early screening and pre-emptive strategies for rare diseases, as well as innovative interventions for conditions like cancer and metabolic disorders, hold significant importance.

P4 (Predictive, Preventive, Personalised and Participatory) medicine aims to reduce the burden of common and rare diseases, manage quality of life, enhance wellness and health- longevity and provide personalised and affordable healthcare solutions. Since India has a diversity of populations with unique socio-cultural and dietary practices and plurality of medical systems the challenges are even more immense in achieving the above objectives. However, integration with traditional knowledge systems like Ayurveda and Yoga have distinct advantages and incorporating these with modern medicine using a common ontological framework could expand treatment options.

Digitally-driven solutions have the potential to enhance precision, scalability, and applicability across healthcare. While omics technologies have propelled precision medicine, their invasiveness and cost remain challenges. AI and IoT-enabled devices are indispensable for data collection and recommendation systems. Collecting an individual's genetic and baseline phenotypic data, alongside non-invasive monitoring in spatio-temporal dimensions, could not only enable predictions of dynamic disease patterns but also help monitor lifestyle and dietary interventions with an individualised and participatory approach. This is only possible in the backdrop of a Data Science driven participatory framework that enables population and individual risk stratification and early actionable precision health interventions. The talk would highlight some key takeaway points during traversing the route from understanding biological intelligence through phenomes, genomes and natural variations and need for getting into the bus for artificial intelligence.

Stringently controlled gene regulation is essential for embryonic development and homeostasis. The complex interplay of transcription factors, cis-regulatory elements and target genes is essential for robust gene regulation. Recently developed of high-throughput technologies have enabled analyses of such complex interplay at the genome-wide scale, providing unprecedented insights into the molecular mechanisms of gene regulation. The high throughput genome-wide approaches include RNA-seq for monitoring global gene expression patterns, ChIP-seq for transcription factor binding and occupancy profiles of histone modifications, ATAC-seq for analysing chromatin accessibility and HiC for mapping the dynamic three-dimensional chromatin conformations. Using the example of zygotic genome activation, I will elaborate the use of these technologies. Further, I will discuss key computational approaches to delineate these elements via integration of multiple large datasets to reconstruct the regulatory networks underlying spatio-temporal gene expression.

Genome-wide association studies (GWAS) is most widely used approach for the genetic dissection of complex traits and has been successfully used in most of the important crops. The present talk would be focused on the basics and principles of GWAS. GWAS is based on linkage disequilibrium (LD), and has several advantages over the linkage mapping including higher mapping resolution, fast and mining of large number of alleles etc. Although GWAS also has certain limitations particularly occurrence of false positives due to population structure, genetic relatedness and multiple testing. We will discuss about why these false positive arises and how false positives can be reduced using different statistical tools. Step by step procedure including data file preparation, statistical steps and their pros and cons will be discussed during the talk, and hands-on training will also be provided for the same using sample data.