actual UNEAK

Some of you may have heard me ramble about my little in house Uneak-like SNP calling approach. I am being converted to using the actual UNEAK pipeline. Why? reason #1 is good god is it fast! I processed 6 lanes raw fastq to snp table in ~1 hour. #2 I am still working on – this will be a comparison of SNP calls between a few methods but I have a good feeling about UNEAK right now.

Here is the UNEAK documentation:

It is a little touchy to get it working so I thought I would post so you can avoid these problems.

First install tassel3 (not tassel 4) you will need java 6 or younger to support it:

To be a hacker turn up the max RAM allocation. Edit to change the line “my $java_mem_max_default = “–Xmx1536m”;” to read something like “my $java_mem_max_default = “–Xmx4g”;” (That is 4g for 4G of RAM).

Go to where you want to do the analysis. This should probably be a new and empty directory. Uneak starts by crawling around looking for fastq files so if the directory has some files you don’t want it using it is going to be unhappy.
Make the directory structure:

../bin/ -fork1 -UCreatWorkingDirPlugin -w /home/greg/project/UNEAK/ -endPlugin -runfork1 

Move your raw (read: not demultiplexed) fastq or qseq to /Illumina/. If you are using fastq files the names need to look like this: 74PEWAAXX_2_fastq.txt NOT like this:74PAWAAXX_2.fastq and probably not a bunch of other ways. It has to be flow cell name underscore lane number understore fastq dot txt. You might find others work but I know this does (and others don’t).
Make a key file. This is as described in the documentation. It does not have to be complete (every location on the plate) or sorted. Put it in /key/

Flowcell        Lane    Barcode Sample  PlateName       Row     Column
74PEWAAXX       1       CTCGCAAG        425218  MyPlate7        A       1
74PEWAAXX       1       TCCGAAG 425230  MyPlate7        B       1
74PEWAAXX       1       TTCAGA  425242  MyPlate7        C       1
74PEWAAXX       1       ATGATG  425254  MyPlate7        D       1

Run it. Barely enough time for coffee.

../bin/tassel3.0_standalone/ -fork1 -UFastqToTagCountPlugin -w /home/greg/project/UNEAK/ -e PstI -endPlugin -runfork1
# -c how many times you need to see a tag for it to be included in the network analysis
../bin/tassel3.0_standalone/ -fork1 -UMergeTaxaTagCountPlugin -w /home/greg/project/UNEAK/ -c 5 -endPlugin -runfork1
# -e is the "error tolerance rate" although it is not clear to me how this stat is generated
../bin/tassel3.0_standalone/ -fork1 -UTagCountToTagPairPlugin -w /home/greg/project/UNEAK/ -e 0.03 -endPlugin -runfork1
../bin/tassel3.0_standalone/ -fork1 -UTagPairToTBTPlugin -w /home/greg/project/UNEAK/ -endPlugin -runfork1
../bin/tassel3.0_standalone/ -fork1 -UTBTToMapInfoPlugin -w /home/greg/project/UNEAK/ -endPlugin -runfork1
#choose minor and max allele freq
../bin/tassel3.0_standalone/ -fork1 -UMapInfoToHapMapPlugin -w /home/greg/project/UNEAK/ -mnMAF 0.01 -mxMAF 0.5 -mnC 0 -mxC 1 -endPlugin -runfork1