Don't follow causal property chains for regulation targets when outputting conventional annotations?

[ukemi]

In gomodel:MGI_MGI_2442120 we have been asked to remove an annotation to '[GO:0043491] phosphatidylinositol 3-kinase/protein kinase B signal transduction' that is being generated as the result of computing over a property chain. The model shows that the gene product, Bank1, is causally_upstream_of_or_within '[GO:0051898] negative regulation of phosphatidylinositol 3-kinase/protein kinase B signal transduction'. The '[GO:0051898] negative regulation of phosphatidylinositol 3-kinase/protein kinase B signal transduction' negatively_regulates '[GO:0043491] phosphatidylinositol 3-kinase/protein kinase B signal transduction' and the regulated process occurs in a B cell.

Two annotations are being generated from this construct:
MGI MGI:2442120 acts_upstream_of GO:0043491 PMID:16546095 ECO:0000316 MGI:MGI:88336 20140701 MGI occurs_in(CL:0000819) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:MGI_MGI_2442120|model-state=production

MGI MGI:2442120 acts_upstream_of_or_within GO:0051898 PMID:16546095 ECO:0000316 MGI:MGI:88336 20140701 MGI negatively_regulates(GO:0043491) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:MGI_MGI_2442120|model-state=production

Although these are both correct. It appears that Users can be confused if they do not pay attention to the qualifiers and maybe we should not output the first annotation???????

ping @LiNiMGI