Screen How-to

`screen` is one of the most useful system tools I've ever used. It allows you to begin a process and then detach from the process to let it continue to run in the background. You can later re-attach to the screen to check progress, even from a remote location.

To install screen on a Debian-based system, you would do:

apt-get install screen

or for a RedHat-based system:

yum install screen

Some usage examples:

To start a job inside a screen, a Ruby script for example, you would issue a command something like this:

> screen my_script.rb

The job begins to run in exactly the same manner if you had not used a screen.

To detach from the screen you would use these keystrokes:

> Ctrl + a, Ctrl + d.

After this the Ruby script is still running.

To reconnect to the running screen you type:

> screen -dr

The 'd' means detach and the 'r' means re-attach to the current terminal.

If you have lots of screens running, you'll be prompted to specify which screen you want to re-attach to your terminal. For example:

> screen -dr
There are several suitable screens on:
5190.pts-0.pluto (03/18/2009 01:36:22 PM) (Detached)
5134.pts-0.pluto (03/18/2009 01:14:08 PM) (Detached)

Type "screen [-d] -r [pid.]tty.host" to resume one of them.

At this point you have to specify like this:

screen -dr 5190.pts-0.pluto

or

screen -dr 5134.pts-0.pluto

I find screen most useful for starting long-running jobs on remote servers. I can start the job, then log out and let it run without any worries of what my local system is doing. I can reboot or log off without any issues. Later I can re-attach the screen to my terminal to check progress as required. When the job is done, so is the screen, they are self-cleaning. :)

More info can be found here:

http://www.linuxmanpages.com/man1/screen.1.php

Noteworthy Blog: The Spittoon

For those of you who attend our computational genetics journal club every other week, you've all heard about this. Say what you will about the "consumer genetics" enterprise, 23andMe maintains an excellent blog. In their "SNPwatch" category, The Spittoon surveys and summarizes the latest findings in human genetics research before they hit the press. About 50% of their content comes from Nature Genetics advance online, and the rest from a smattering of other journals. They usually offer a one-page summary of the research findings detailing the associated SNP's rs-number, risk allele, odds ratio estimate, and the sample size used.

The Spitoon

What's the Right Analysis for this Data?

If you've ever had trouble getting started doing a data analysis, you are certainly not alone. Should I run an ANOVA, MANOVA, ANCOVA, or MANCOVA? Should that have been a McNemar's test, a Kruskal-Wallis, or a Mann-Whitney U?

To at least pin down the statistical test you should run, consult these excellent flowcharts by Marylyn Ritchie, Jason Moore, and Tricia Thornton-Wells. They have been stuck to my wall for years, and whenever I have a new type of data to look at, I consult them to get headed in the right direction!

Will

UPDATE 2009-04-06: Be sure to check out the follow up to this post, What's the Right Analysis part II, for a link to example applications of these methods with the appropriate Stata code.

SNPper: quickly get info about a list of SNPs

Many of you have used this before, but for those who haven't, SNPper is a convenient little web application for quickly annotating results. So you've done your association analysis and have a list of rs-numbers you'd like to quickly get more information about. You could always look these up on the UCSC genome browser or in dbSNP, but if you have hundreds and you want some quick info such as what genes they're in, what genomic position they occupy, whether they're exonic, nonsynonymous, in a promoter region, etc., try SNPper. You can register for free, or use their guest access which doesn't require registration. Paste a list of rs-numbers into their box, and click find. It also gives you some nice options for exporting the information to a file. They also offer other bioinformatics tools, such as gene ontology browsing and transcription factor binding site identification.

SNPper

Linux Tutorial

Last week I posted a one-page reference guide that gives a short description of the Linux commands I most commonly use. To accompany this, here is a detailed walkthrough filled with examples that will introduce any beginner to the basics of using Linux in just a few hours. Many thanks to Eric Torstenson in the Ritchie lab for putting this together.

Linux Tutorial (PDF)

Linux Command Line Cheat Sheet

Whenever we have new students rotating through our lab who've never used Linux I always end up scrounging around the world wide series-of-tubes only to find some command line reference that's not really useful for students. So I made my own one-page reference guide that gives a basic description of 99% of the commands most of you will use. Feel free to distribute.

Linux / Unix Command Line Cheat Sheet (PDF)

Pubmed Searches as an RSS feed

As Stephen nicely posted earlier, RSS feeds are a very powerful way to keep up with the literature -- they "push" the information to you. In addition to subscribing to individual journals, you can subscribe to a PubMed search! This will let you keep up with ALL PubMed indexed journals.

To subscribe to a PubMed search, first go to www.pubmed.org and enter your search terms. Once you retrieve a search listing, you'll see a bar that says

Display Summary Show 20 Sort By Send to

The SEND TO drop down box will allow you to select an RSS Feed. Once you select this, you'll be taken to a page with a button that says "Create Feed". When you click this, you'll get a new page with a little orange XML button. Click it and your browser will give you the option to subscribe to the feed. Once you subscribe, there are lots of ways to read RSS Feeds, which we'll probably get to in another post.

Enjoy!