search

Illumina / ExpansionHunterDenovo

A suite of tools for detecting expansions of short tandem repeats
Other
79 stars 25 forks source link

Set seed for reproducible resampling results #3

Closed katherinesmith closed 4 years ago

katherinesmith commented 4 years ago

Currently the results of outlier analysis can vary from run to run for both the locus and motif based analysis. More specifically, top_case_zscore can vary and this sometimes leads to different numbers of lines being written to the output .tsv.

I believe that this is due to the use of numpy random.choice() without setting numpy random.seed(). It would be great to set a seed so that results can be reproduced exactly.

egor-dolzhenko commented 4 years ago

Thank you for noticing and reporting the problem. We will fix this shortly.

egor-dolzhenko commented 4 years ago

I am sorry this took awhile! The v0.8.1 release should fix the issue.

katherinesmith commented 4 years ago

Thank you @egor-dolzhenko! Classic choice of seed :)