Transcriptome assembly notes

ApTranscriptome

Script to run diginorm on interleaved files failed due to error:

Error: Improperly interleaved pairs HWI-ST1073:326:D25DAACXX:6:1101:1359:1925 HWI-ST1073:326:D25DAACXX:6:1101:1359:1925

making hashtable

This is an outstanding issue with khmer.

PYTHONPATH status not remembered between sessions so need to set in script.

Modified /khmer/sandbox/interleave.py script to add Illumina 1.3 fastq format read tag (‘/1’ or ‘/2’) to sequence names

#! /usr/bin/env python
import screed, sys, itertools

srfile = sys.argv[1]

n = 1

for read in screed.open(srfile):
    name = read.name 
    if n%2!=0:
        name += '/1'
    else:
        name += '/2'
    n += 1
    print '@%s\n%s\n+\n%s' % (name,
                              read.sequence,
                              read.accuracy,
                             )

First useful python script! Not much, but have to start somewhere. Saved as gist

Scripts are working! Confirmed that the –paired option discards both reads of paired group by running grep -c @HWI.*/1 A22-00.notCombined.fastq.out.keep and confirming that the same number as grep -c @HWI.*/2 A22-00.notCombined.fastq.out.keep. Note that I first ran this with the pattern only /1 and this did not result in the same number of reads as it also matched at some places in the quality scores.

Problem in that final normalization is across both paired and unpaired fastq files. Could use strip-and-split-for-assembly.py or after recommendation from C. Titus Brown on khmer list, switch order of workflow to

normalize reads using the -p option first
then merge reads using FLASH
run velvet with the extended and notCombined files output from FLASH

Huh. Turns out that Trinity can also do in silico normalization

Got XSEDE account for mason so can try this out!

ApGXL

Email from CP that worrying levels of mortality in 26C Phytotron. Discussed removing the daily ramp of 3C (23-29C in the 26C chamber) but this could cause problems with CP’s work as ants from multiple projects are in the same phytotron chambers. Availability of other chambers? Otherwise move to the 23C (20-26C) chamber. Will still experience chronically warmer temps, even if there is overlap with the 20C chamber.

Computing

Wow. Did a dumb thing while setting up ssh access for my github repository on antlab. Accidentally set origin to wrong repository and after a pull ended up with lots of commits files that did not belong! Yike.

Stackoverflow yet again came to the rescue.

To undo the last git pull use git reset HEAD@{1}.

Also note that git reflog can be useful as it shows the history of HEAD

https://help.github.com/articles/changing-a-remote-s-url

More git fun. Started working on a development ‘transcriptome’ branch that pipes together all my working scripts, but breaks everything at this point. Exactly what branches are for! So did

git checkout -b transcriptome

Made all the changes I wanted. Commited them.

Then changed back to master git checkout master

But I actually would like to include updates to the README file in my master branch. Google yet again gave the answer and it’s quite simple

jstantongeddes@AntLab:~/climate-cascade$ git branch
* master
transcriptome
jstantongeddes@AntLab:~/climate-cascade$ git checkout transcriptome projects/ApTranscriptome/scripts/README.txt
jstantongeddes@AntLab:~/climate-cascade$ git status
# On branch master
# Changes to be committed:
#   (use "git reset HEAD <file>..." to unstage)
#
#   modified:   projects/ApTranscriptome/scripts/README.txt
#
# Untracked files:
#   (use "git add <file>..." to include in what will be committed)
#
#   projects/ApTranscriptome/results/01-trimclip/
#   projects/ApTranscriptome/results/A22-velvet_61/
jstantongeddes@AntLab:~/climate-cascade$ git commit -m "Update to README.txt in /projects/ApTranscriptome/scripts"
[master e1e2259] Update to README.txt in /projects/ApTranscriptome/scripts
 1 file changed, 20 insertions(+), 1 deletion(-)
 rewrite projects/ApTranscriptome/scripts/README.txt (100%)

And that’s it! Changes from the README file on my transcriptome branch are now in master.