Amazon Lightsail now offers servers at $3.5 per month with 1 vCPU, 512MB of memory, and 1Tb of data transfer. Similar services are available at DigitalOcean. We can host personal VPN (virtual private network) servers cheaply. Docker containers enable quick configuration of these servers, even for people with little Linux experience.
As someone who uses Google search and Google Scholar dozens times a day, I am especially concerned by the blockage of Google websites for scientists in some countries.
During the preparation of my NIH proposal in May 2018, I started a OneNote document to organize the information I gathered. These are my notes and some tips. Hope it can be useful to new investigators. I got lucky with NIH only once, so I am not that “seasoned”. This is based on my personal opinions, experiences, and lessons.
1. Suck it up and write it I use one Microsoft Word document to write all the components: Project Summary, Narrative, Specific Aims, Research Strategy, References, Facilities, Equipment, Personnel justification, Resource Sharing Plan, etc.
iDEP is a tool for interactive analysis of RNA-Seq data. After analyzing their data, users have the option to download a R Markdown file, toegether with other data files, to reproduce the entire workflow. This blog is both the easiest and the hardest of all. It is easy because everything below was automatically generated by iDEP; it is hard as I have to write code to write code when developing iDEP.
Press release
Thirteen high school students from across the state participated in this year’s Data Analytics Camp at South Dakota State University (SDSU) main campus from June 18th to 21st, 2018. There were six girls and seven boys. The students lived in college student’s dorms for four days. Most learning activities were in a computer lab in Solberg Hall. As most college students are away during the summer, the high school students got to use a lot of facilities and services.
As a scientist who also suffers from vitiligo, I spent a day trying to read the current SCIENTIFIC findings published in peer-reviewed scientific journals.
Two type of vitiligo: segmental (localized) vs. non-segmental. Most are non-segmental. Some evidence suggests that vitiligo patients are low in antioxidants, vitamin D, zinc, and copper. Vitiligo patients often suffer low glutathione peroxidase (GPx) as suggested by this and other papers. Vitiligo is a loss of pigmentation in the skin.
Spot instances on Amazon Elastic Compute Cloud (EC2) allow researchers to do high performance computing at very low cost. For example, a 64-core workstation with 256 GB of memory can be rented at about $0.7 per hour, which is ~80% off regular on-demand prices. You can request 10 of such servers and use them for 5 days and 2 hours and it only cost you a few hundred dollars.These are spare computing capacities that could be abruptly terminated.
This website is created using RStudio following this tutorial with tips from Ming Tang.
To start a new post, choose “New post” from Addins. Use Markdown for texts. Use RMarkdown if it contains R markdown to render plots, etc.
Choose which files you want to work on from the right bottom window. Save the files changed and the webpage is rendered automatically in the Viewer window.
To disable a widget, change the header from active=true to active=false
Update. 11/13/17 Thanks to Mike Love for pointing out an error. This is just my understanding. There maybe other errors.
To allow complex study designs in iDEP(http://ge-lab.org/idep/), I tried to understand how factoral models are built and the desired results are extracted from DESeq2. The following is based on the help document from the resutls( ) function in DESeq2, plus some of Mike Love’s answers to questions from users.
With the help of my students, I finally figured out how to use Docker to define an isolated environment. I wanted to run the most recent version of R and Bioconductor in a Linux machine without bothering IT support people.
Turns out it is a one-liner, if you have the docker software installed:
sudo docker run --name=my-r-base --rm -ti -v /home/gex/Rdocker:/myData bioconductor/release_core2 /bin/bash It downloads the most recent Bioconductor image from Docker hub, and runs it.
Using the mass shooting data from Mother Jones website, I tried to forecast the number of mass shoottings and total fatalities in the rest of 2018. Some of the R scripts are from this website. Please let me know if I there is any error. I am not an expert on time series analysis. Twitter @StevenXGe
In recent years, mass shootings are becomming more frequent and more deadly. The models suggest for the rest of 2018, there probably be about 6 more shootings, killing about 59 people in the U.