European Conference on Computational Biology (ECCB12) - Day 1

We're back at ECCB12, don't worry I didn't forget. :)

Name: Zsuzsanna Sukosd

Date: 9th to 12th September 2012

Location: Basel, Switzerland

Website: http://www.eccb12.org/

The first day of the main conference continued the workshops and tutorials, covering a wide range of subject areas. I attended a very interesting tutorial on inferring genetic diversity from next-generation sequencing data. Two examples were covered in detail throughout the lectures: HIV and cancer. In both cases, estimating the genetic diversity of samples is highly relevant for clinical purposes. With the emergence of second- and third-generation sequencing technologies, one might naively expect that assessing genetic diversity has mostly become an experimental exercise. It turns out the problem is not nearly that easy.

In the tutorial, we were presented with a simple calculation. Let's assume a sequencing error rate of 0.1% per base pair (standard for the Roche 454 sequencing technique) and an average read length of 500 bp; this means that we will observe incorrect reads in about 40% of sequencing reads. This might not be a big problem when our samples are relatively uniform as for most humans. HIV-1 is not very uniform and is reported to have a mutation rate of around 3 in every 100 000 nucleotides. The HIV genome size is around 10 000 nt, so for every 10 new viruses one will be new and different. Given that an HIV-positive patient produces between 10¹⁰ and 10¹² new viruses every day, every patient may produce billions of variant HIV viruses each day. Suddenly we are faced with a big problem. If we see a difference in sequencing reads, is it because there has been a true mutation in the virus being studied, or is it just a sequencing error? And because the reads from sequencing technologies are rather short, even if we assume that the mutations were genuine, how do we know which reads fit together into true virus genome sequences?

So there is clearly a need for computational and statistical methods for error correction and genome reconstruction. The tutorial was very successful at introducing these theoretical problems and exploring existing bioinformatics and computational biology solutions.

In the evening, the main conference started with an opening talk, keynote lectures and highlight presentations, followed by a reception in the impressive setting of a huge local church. To give an idea of the scale of the conference, there were around 1200 attendees - they had to install a custom-made WiFi system in the conference center so we could all be on the internet at the same time, and they even managed to arrange special additional trams in the city just to transport the conference participants to the reception!

(Conference logo from website)