resources
visualizations
presentations
lunch break
Distractions and amusements, with a sandwich and coffee.
genomics + data mining
ICDM2012 Keynote
Needles in Stacks of Needles: genomics + data mining
visual abstract
abstract
In 2001, the first human genome sequence was published. Now, just over 10 years later, we capable of sequencing a genome in just a few days. Massive parallel sequencing projects now make it possible to study the cancers of thousands of individuals. New data mining approaches are required to robustly interrogate the data for causal relationships among the inherently noisy biology. How does one identify genetic changes that are specific and causal to a disease within the rich variation that is either natural or merely correlated? The problem is one of finding a needle in a stack of needles. I will provide a non-specialist introduction to data mining methods and challenges in genomics, with a focus on the role visualization plays in the exploration of the underlying data.
references
The title of the talk was drawn from the paper
Gregory M. Cooper & Jay Shendure Needles in stacks of needles: finding disease-causal variants in a wealth of genomic data Nature Reviews Genetics 12, 628-640 (September 2011)
I will be posting a full list of references for the talk shortly.
news + thoughts
Using Circos in Galaxy Australia Workshop
A workshop in using the Circos Galaxy wrapper by Rasche and Hiltemann. Event organized by Australian Biocommons.
Galaxy wrapper training materials, Saskia Hiltemann, Helena Rasche, 2020 Visualisation with Circos (Galaxy Training Materials).
Essence of Data Visualization in Bioinformatics Webinar
My webinar on fundamental concepts in data visualization and visual communication of scientific data and concepts. Event organized by Australian Biocommons.
Markov models — training and evaluation of hidden Markov models
With one eye you are looking at the outside world, while with the other you are looking within yourself.
—Amedeo Modigliani
Following up with our Markov Chain column and Hidden Markov model column, this month we look at how Markov models are trained using the example of biased coin.
We introduce the concepts of forward and backward probabilities and explicitly show how they are calculated in the training process using the Baum-Welch algorithm. We also discuss the value of ensemble models and the use of pseudocounts for cases where rare observations are expected but not necessarily seen.
Grewal, J., Krzywinski, M. & Altman, N. (2019) Points of significance: Markov models — training and evaluation of hidden Markov models. Nature Methods 17:121–122.
Background reading
Altman, N. & Krzywinski, M. (2019) Points of significance: Hidden Markov models. Nature Methods 16:795–796.
Altman, N. & Krzywinski, M. (2019) Points of significance: Markov Chains. Nature Methods 16:663–664.
Genome Sciences Center 20th Anniversary Clothing, Music, Drinks and Art
Science. Timeliness. Respect.
Read about the design of the clothing, music, drinks and art for the Genome Sciences Center 20th Anniversary Celebration, held on 15 November 2019.
As part of the celebration and with the help of our engineering team, we framed 48 flow cells from the lab.
Each flow cell was accompanied by an interpretive plaque explaining the technology behind the flow cell and the sample information and sequence content.
Scientific data visualization: Aesthetic for diagrammatic clarity
The scientific process works because all its output is empirically constrained.
My chapter from The Aesthetics of Scientific Data Representation, More than Pretty Pictures, in which I discuss the principles of data visualization and connect them to the concept of "quality" introduced by Robert Pirsig in Zen and the Art of Motorcycle Maintenance.