Daniel:Notebook/Haplotyping/SISSORPipeline/2014-12-3
Segmentation Analysis Pipeline (Started 12/2/2014)[edit]
Reworking the Analysis Pipeline[edit]
Looks like there are many steps to rework the pipeline. I'll start at the beginning.
Step 1-Python Script[edit]
This doesn't actually need reworking, but I'm having trouble running the command.
Command
python varbin.50k.sam.py pgp22_chamber1.BQSR.realigned.sam test.varbin50k.txt test.varbin50k.stats.txt
Result
Traceback (most recent call last): File "varbin.50k.sam.py", line 129, in <module> main() File "varbin.50k.sam.py", line 71, in main thisRatio = float(binCounts[i]) / (float(counter) / float(len(bins))) ZeroDivisionError: float division by zero
I think the problem is the chromosomes in the script are named "chr1","chr2",etc., whereas the chromosomes in my file are named "1","2",etc. This is a common problem I have run in to, and should be addressed. This is an example command to append "chr" to the 3rd column (reference name column) of the output file.
awk 'BEGIN {OFS = "\t"} $3 = "chr" $3' infile.sam > outfile.sam
Repeating the first command now produces a viable result.
Step 2-Cutoff To Define Initial States[edit]
Eric's original pipeline called for anything less than 20 read count to become state 1, and anything 20 or greater as state 2. This seems like a pretty arbitrary cutoff, so I'm going to explore a little bit to see why, and maybe whether I can improve on that version.
I took the data set I already analyzed (PGP1-22; chamber 1) and sent it to MATLAB.