Team 4

Identifying and visualizing outliers in multi-variate summary statistics space

Members:

• Katie Lotterhos (team lead)
• Sara Schaal
• Daren Card
• Liuyang Wang
• Caity Collins
• Bob Verity

Goals

• goals go here
• To create a package for summarizing multivariate data that is a common product in population genetics analyses.
• To construct a Shiny-based visualization medium for multivariate datasets.

Status

Day 2:

• Collecting problems that need multi-variate summaries. Have several from people here at the event.
• Feedback sought: data sets and problems needing Fst statistics. Especially sequence data.
• Looking into Mahalanobis distance-methods, including the one implemented in hclust in R.
• Created Github repo and package skeleton.

Day 3:

• Have now real and simulated datasets from which sites under selection and neutral sites can be drawn.
• Have a working connection to shiny app.
• Facing the challenge that the statistics from different methods can have vastly different variances.
• Working on plots that can be reproduced in shiny.

Day 4:

• Working on statistics angle of multivariate summary.
• Lots of progress with Shiny app. Challenges with circular dependencies.
• Some nice charts and plots in the works. Aiming to use ggplot2, however that makes huge objects by storing data internally. No easy solutions out there.

Products

Status: We have a Github repository and skeleton package complete.

• R package "MINOTAUR" (still under development)