Trends in Bioinformatics

Proof of Concept

1. Overview
2. General Visualizations
3. Visualization Maps
4. Bigram Relationships
5. Sankey Diagrams

Project Goals

Graphic Visualization of trends in Bioinformatics

1. Usage of bioinformatic tools and technqiues (ex. sequence alignment, genome assembly, metagenomics etc)
2. Relationship of tool development within analytical pipelines
3. Geographic hotspots for development of bioinformatic tools and techniques

RLadies Workshop

1. Racing Bar Graph and Sankey Diagrams (slides)
   • Live-coding worksheet (complete version here)
2. Visualization Maps and Bigram Relationships (slides)
   • Live-coding worksheet (complete version here)

Team

• Diana Lin
• Emma Garlock
• Jasmine Lai
• Raissa Philibert
• Morgana Xu
• Lucia Darrow
• Swapna Menon
• Shannon Lo
• Elliot YKF

Approach

1. Webscraping using the package fulltext package, processing the data in the package pubchunks code for this can be found in R/Fulltext_Workflow.Rmd.
   • We will be looking for: doi, Year of Publication, Journal Name, publisher, author, research institution, title, abstract.
   • The keywords used for searching determined by terms in the EDAM ontology database and previous knowledge of the field.
   • The keywords have been set up in raw-data/SearchTerms.csv. We are working on optimizing the R/Fulltext_Workflow.Rmd file to loop through this CSV.
   • Due to access issues, we are only looking at papers in plos
2. Visualization of trends in the the data with ggplot and other R visualization packages

Pre-requisites

These are the R Packages needed:

• fulltext
• pubchunks
• tidyverse
• magrittr
• dplyr
• purrr
• here
• future
• lubridate
• stringr
• maps
• viridis
• rgeos
• sf
• ggmap
• maptools
• igraph
• ggraph
• tm
• gganimate
• data.table
• textrank
• udpipe
• tidytext
• ggplot2
• magriter
• plotly
• googleVis
• ggrepel
• egg
• grid
• ggalluvial
• widyr
• readr
• tidygraph

These are the Python (python3) packages needed:

• googlemaps
• pandas
• Will also require a geocoding API from Google

Usage

Webscraping

To webscrape, run database_parallel2.R in the R directory after modifying the topic and output filename based on the names in the raw-data/SearchTerms.csv.

Visualization

Three different types of information will be plotted

1. General summary plots of topic coverage over time and topic coverage based on journal.
2. Analysis of bigrams (combinations of two words) used in publications to see common phrases and their relationships. Also looking to connect the authors and their subject matters using a sankey diagram.
3. Geographic mapping of insitutions involved in bioinformatic research. The file being used to generate visualizations R/visualization_map.Rmd
   • Dynamic heatmaps generated with colours based on the number of active insitutions in the area. The maps provide global view, as well as specific USA and European maps due to the high number of results in those areas. Hovering over the countries/states shows the number of institutions and hovering over the points (where shown) identifies the institutions.
   • An interactive world map showing the most used keywords in each country. Hovering over the countries shows the topic names.