View Single Post
Old 01-12-2017, 06:33 AM   #42
Location: Germany

Join Date: Oct 2014
Posts: 14
Default A question reagarding the partitioning option of BBnorm

I want to preferentially assemble the genome of a low abundant community member from a metagenome, so I am interested in the partitioning option of BBnorm.

I have some questions on how to choose the best parameters though:

-for the other bbnorm workflows (normalization, filtering, error correction) you recommend the "prefilter" option. Is this also recommendable for the partitioning workflow? (Because this option is used in most of the example-usages of BBnorm in the documentation EXCEPT the partitioning workflow)

-from the description, I assumed that by giving "outlow, outmid and outhigh" arguments, the usual normalization workflow would be overridden and ALL reads would be grouped into one of these categories. However the preliminary output of BBnorm states that a "target depth" of 100 and a "min depth" of 5 is being applied. Does that mean that all reads below a coverage of five will be discarded? Do I need to adjust the "mindepth" parameter as well?

-Our job-submission pipeline requires the specification of a maximum RAM usage for all scripts started. However bbnorm keeps exceeding this value (which leads to a termination of the job). I kept increasing the memory limit of BBnorm using the "-Xmx" argument upto 200G, but always bbNorm exceeds the alloted limit (even if using the "prefilter" option above).
Do I have consider any additional memory requirements of the script, in addition to the "-Xmx" limit? How would I determine how much memory is needed?
(The dataset consists of about 84.547.019 read-pairs calculated a "Cardiality" of 5.373.179.884, but I do not know how exactly to interpret this value).

Thanks for any suggestions.
jov14 is offline   Reply With Quote