question about coverage

[kkamii]

hi

I made two paired end fastq file with different seed option (-rng), and same coverage option (-c).
one fastq file have mutation (frequency 100%) and the other have reference allele.

and then, I mixed fastq file to make frequency 50%

but when I check the bam file, coverage is so different that frequency is not 50% (about 8%)

how can I fix it??

[zstephens]

Greetings,

I tested this out using the following commands:

1) generate 100x coverage, no variants
python genReads.py -R 101 -r chr1_subset.fa -p 1 -M 0 --rng 123 -o out1 --bam -c 100

2) generate 100x coverage, with 2% random variants
python genReads.py -R 101 -r chr1_subset.fa -p 1 -M 0.05 --rng 456 -o out2 --bam -c 100

3) combine datasets
python mergeJobs.py -i out1 out2 -o mix -s /path/to/samtools --no-job

And the resultant data had a 50-50 split as expected:

(note that I just pushed some QoL updates to mergeJobs.py, including the "--no-job" option).

It's possible that you got really bad luck in the RNG and between the two runs there was a substantial difference in what position the reads were sampled from. In order to rule out gc-bias, you could try adding the option --gc-model models/gcBias_uniform.p to the simulation commands.

-Zach