FAQ

Index

1. Which chunk size should I choose in the 'configuration file' for the imputation step? 2. GUIDANCE did not finish successfully since I see temporal files in the working directory. How can I check what went wrong? 3. Some of my imputation summary files did not finish and get stuck in the 'setting switch rates' step. What went wrong? 4. GUIDANCE failed when making the QQ plot and the Manhattan plot. What is the problem?

Answers

1. Which chunk size should I choose in the 'configuration file' for the imputation step?

IMPUTE2 authors recommend splitting each chromosome into chunks of a few megabases (Mb). Hence, taking advantage of imputing small chunks in parallel, the real computing time and memory usage decrease without compromising the accuracy. Making the choice of the best chunk size to run the workflow mostly depends on available computational resources, and computational needs depend on sample size and reference panels. GUIDANCE allows users to decide the suitable chunk size accordingly with their particular needs through the configuration file. See the figures below based on our experience with different chunk size when imputing with different reference panels.

2. GUIDANCE did not finish successfully since I see temporal files. How can I check what went wrong?

Check the logs, error and out files for GUIDANCE execution. In 'Jobs' directory, if some particular job failed, check the error files from those that have been 'RESCHEDULED' or 'RESUBMITTED' for more details.

3. Some of my imputation summary files do not finish and get stuck in the 'setting switch rates' step. What went wrong?

IMPUTE2 gets usually stuck at this step when not enough memory is assigned to this job without reporting a memory error. This can be solve re-launching the GUIDANCE execution after modifying the constraints increasing the memory, or diminishing the chunk_size_analysis in the configuration file.

4. GUIDANCE failed when making the QQ plot and the Manhattan plot. What is the problem?

First of all, check if there are some outputs after the SNPTEST. If the answer is 'no' make sure than the sample file was in the correct format. Commonly, this occurs because the sample file was not the adequate and there are no results from the SNPTEST to make the plots. This error can also indicate that the imputation did not finish due to memory requirements, that results in an error when running SNPTEST that makes impossible to generate the plots. If this is the problem, see how to modify the constraints to optimize the resources so as to run the program properly. This problem can also be caused by an inappropriate installation of the 'gap' or/and 'sfsmisc' R packages.