APBioNET Event

Finding Antimicrobial peptides in the global microbiome using machine learning

Speaker: 

Luis Pedro Coelho is a group leader at the Centre for Microbiome Research at the Queensland University of Technology. His research focuses on using very large scale datasets of the global microbiome to understand microbial ecology. His group is also known for developing high-quality tools, most notably SemiBin for metagenomics binning. Before moving to Australia, Luis got a PhD from Carnegie Mellon University in the (US), worked at the EMBL in Germany, and at Fudan University in China.

Abstract:

Antimicrobial peptides (AMPs) are small peptides (operationally defined as those up to 100 amino acids) which kill or inhibit microbes. AMPs are produced by organisms from all domains of life, including by bacteria (which use it to compete with each other). They are of interest for drug development as they are less likely to lead to resistance than traditional antibiotics. However, the vast majority of AMPs are unknown. We have developed a machine learning approach to predict AMPs from metagenomic data. We have applied this approach to the global microbiome and found nearly one million novel AMPs. We tested 100 in vitro and found that 79 had antimicrobial activity. Subsequently, we tested the top candidates in vivo in a mouse model of infection and found that they were effective in reducing bacterial load at a level comparable to polymyxin B a clinically used antibiotic. This work demonstrates the power of machine learning to discover novel bioactive molecules from the global microbiome.

LINK

Workshop on Next Generation Tools: Exploring Bioinformatics with Julia and Rust

Speaker:

Dr. Ragothaman M. Yennamalli is a computational biologist at SASTRA Deemed to be University at Thanjavur, Taml Nadu. He has more than a decade of experience in predictive modelling and biomolecular simulation projects. Dr. Yennamalli’s skills involve machine learning, systems biology, molecular docking, molecular dynamics simulation.

Abstract:

The rapid growth of biological research has led to the availability of an overwhelming amount of data. In this landscape, where the scale and complexity of biological data continues to grow exponentially, the need for robust, secure, and efficient methods to extract meaningful insights becomes increasingly critical. In the past, Bioinformatics tools primarily relied on conventional programming languages for data analysis, often capped by limitations in scalability and speed when processing vast datasets. However, the advent of modern languages such as Julia, with its prowess in high-performance computing, and Rust, renowned for its focus on safety and system-level programming, present themselves as formidable contenders in addressing the escalating complexities of biological data. The in-built parallel processing capabilities, coupled with the emphasis on memory safety and performance help researchers in producing significant discoveries while reducing the risk of potential errors inherent in handling complex biological information. Moreover, the versatility of Julia and Rust extends beyond their individual strengths. Their interoperability and potential to integrate with existing Bioinformatics tools and libraries further augment their utility in the field. This allows for the construction of comprehensive and robust pipelines for genomic, proteomic, and metabolomic analyses. A growing community of programmers dedicated to developing tools tailored for Bioinformatics applications proves to be the driving force. The dynamic synergy between efficient languages such as Julia and Rust in finding solutions to the ever increasing demands of data-analysis brings in the “Next-Generation” of Bioinformatics. Their speed, robustness, and versatility present them as transformative tools that enable researchers to catalyze the advancements in biological research and unravel the fundamental mechanisms of life.

Link

APBioNET Talks: Dynamic modeling of chromosomal instability in somatic genomes

Abstract:

Chromosomal instability (CIN), a constantly high frequency of chromosome segregation errors during cell divisions, is a major form of genome instability and plays an import role in intra-tumour heterogeneity, metastasis, and therapy resistance. CIN often leads to structural or numerical chromosomal alterations, such as structural variants and copy number alterations. Linking these alterations detected from cancer genomics data with stochastic modelling and Bayesian inference provides a powerful approach to quantify CIN in an evolutionary context, which helps to better understand cancer evolution and inform cancer treatment. In this talk, I will share our work on modelling experimental and real data with this approach.

Speaker Profile:

“I am currently a Surrey Future Fellow, at Section of Systems Biology and Surrey Institute for People-Centred AI, University of Surrey.
Before joining Surrey, I was a Postdoc at Department of Cell and Developmental Biology University College London, where I worked on dynamical modelling of chromosomal instability (CIN) in cancer genomes. Previously, I was a Postdoctoral Fellow at Genome Institute of Singapore, where I mainly developed pipelines and methods to analyse tumour heterogeneity and clonal evolution in liver and lung cancer genomes. I completed my PhD in Computational Biology at School of Computing National University of Singapore, where I developed machine learning and phylogenetic methods for problems related to lateral gene transfer. I obtained my Master’s and Bachelor’s degree from Software Engineering Institute East China Normal University, where I led the development of platforms for high-throughput biological data analysis, including RNA-Seq and proteomic data.

My research is in the broad field of computational biology, which bridges software engineering, machine learning, algorithms, statistics, phylogenetics, population genetics, and omics. I am particularly interested in developing new computational methods and models to address important biological problems related to human health. My goal is to facilitate the mining of new knowledge from the accumulating huge amounts of data for the biological and biomedical community. I have developed several new methods and applied available methods to tackle basic questions arising in the study of species and cancer evolution. My current primary interests are evolutionary dynamics of cancer genomes, especially those driven by CIN, which are still less well studied than point mutations but critical in tumorigenesis and patient treatment.
” https://icelu.github.io

REGISTER NOW

APBioNETTalks 8: Everything Everywhere All at Once: Lytic Polysaccharide Monooxygenases – Structure, Function, and Dynamics

Abstract:
Cellulases are used by the biofuel industry for conversion of cellulose into glucose, which can be converted into bioethanol. While 1st generation biomass (starch and sugarcane) are currently established routes for production of bioethanol, the use of 2nd generation biomass (consisting of lignocellulose) is still in its infancy in India. Deconstruction of lignocellulose is a challenge that is overcome by using Celluases, a broad term for enzymes that act on Cellulose. Our focus was on two industrially important cellulases: Endoglucanases and Lytic Polysaccharide Monooxygenases (LPMO). While endoglucanase cleaves amorphous cellulose, LPMO acts on crystalline cellulose polymers.
The talk will focus on using diverse computational research (such as Elastic Network Models, Evolutionary Analysis, Machine Learning, and RNA-Seq analysis) that were performed to understand their thermostability and structural dynamics. Specifically, elastic network models (ENM) revealed the correlation of their function with respect to thermostability in Endoglucanases, and ENM highlighted the structural dynamics of the substrate binding region of LPMOs. Similarly, evolutionary analysis of the catalytic domain of LPMOs showed the substrate specificity (cellulolytic or chitinolytic) and the implication in the AA10 family of LPMOs. Recently, we developed a machine learning-based functional annotation tool for classifying a sequence belonging to either AA9 or AA10 family of LPMOs. Currently, at SASTRA university my lab is focused on developing a knowledge base for LPMOs that implements Big Data analytics along with natural language processing.

Speaker Bio:
Dr. Ragothaman M. Yennamalli is a computational biologist at SASTRA Deemed to be University at Thanjavur, Taml Nadu. He has more than a decade of experience in predictive modelling and biomolecular simulation projects. Dr. Yennamalli’s skills involve machine learning, systems biology, molecular docking, molecular dynamics simulation

More info:

https://apbtalks.apbionet.org/event/everything-everywhere-all-at-once-lytic-polysaccharide-monooxygenases-structure-function-and-dynamics/