Seminars

|

Do you want to attend a seminar?

Apply for a seminar

Seminars


Seminar
New computational tools for the analysis of microbiome dynamics – Eran Halperin
Tue 25 Jun 2019 11:00 am Institut Pasteur Auditorium Jaques Monod – MONOD (66)

High-throughput microbiome analysis has become ubiquitous over the last few years. However, the interpretation of the data is often non-trivial and highly depends on the methodology used for the analysis. I will describe a few methods for the analysis of microbiome in contexts that are typical in such analyses. First, I will describe a new method for microbial source tracking, that is, finding the sources of a microbiome sample. I will demonstrate how using the method one can reach very different conclusions (that make more biological sense) than using previous methods. Specifically, I will show examples on the dynamics of gut microbiome in babies, and on the usage of source tracking as a tool for disease diagnosis. Second, I will discuss novel approaches for the analysis of time-series micorbiome data, and here too, I will show how this new approach results in new biological insights, particularly on the dependence of microbiome in the future on the current composition of the microbiome. The talk will be self contained, and I will not assume any expert knowledge in computer science or statistics.



Seminar
Evolution of information in HIV-1 protease – Chris Adami
Thu 6 Jun 2019 02:00 pm Institut Pasteur Auditorium Francois Jacob – BIME (26)

Highly-active anti-retroviral therapy has been extremely effective at maintaining low levels of viral load in HIV-infected individuals, but emerging drug resistance is threatening those gains. When therapy is interrupted even briefly, HIV can evolve resistance to one or multiple drugs. Understanding how to stop viral evolution is an important goal of current research. I use HIV-1 protease sequences from public databases to study the dynamics of evolution over a span of nearly ten years, to compare patterns of adaptation in populations that are drug-naive to those that have taken one or multiple protease inhibitors. Using information theory, I show that the amount of information stored in protease sequences of patients that are on drug therapy has been increasing over time, suggesting that they are adapting to the drugs. In comparison, there is no increase in information in the sequences of patients that are drug naive. However, for the virus the increase in information comes at a price: because most of the information is stored in correlations between residues, the sequences are evolving into a more rugged area of the fitness landscape, which could make further evolution more difficult. While the data up to 2006 do not suggest a slowing down of evolution, such a trend may exist in data from later years not analyzed here.