When the output of annotate changes (due to a change in the sensitivity of the search, for instance), the marker-classification and nn-classification will skill skip some steps.
In marker-classification this will cause some features to be incompatible with the actual gene annotations (e.g., marker frequency remains the same, when it should have changed). In both marker-classification and nn-classification, the provirus outputs will remain intact, even if no provirus was detected in the second execution (leading to an error in summary, as exemplified below).
Reproducing the bug
Run the end-to-end module twice to classify LC735414.1, first with -s 4.2 and then with -s 1. A provirus will be detected when running with -s 4.2 but not with -s 1, causing a bug in the summary module.
When the output of
annotate
changes (due to a change in the sensitivity of the search, for instance), themarker-classification
andnn-classification
will skill skip some steps.In
marker-classification
this will cause some features to be incompatible with the actual gene annotations (e.g., marker frequency remains the same, when it should have changed). In bothmarker-classification
andnn-classification
, the provirus outputs will remain intact, even if no provirus was detected in the second execution (leading to an error insummary
, as exemplified below).Reproducing the bug
Run the end-to-end module twice to classify LC735414.1, first with
-s 4.2
and then with-s 1
. A provirus will be detected when running with-s 4.2
but not with-s 1
, causing a bug in thesummary
module.