PostprocessGermlineCNVCalls outputs wrong genotyped-segments results

[DiabloRex]

We have a partial duplication on chr19 (at 0 ~ 30M) in mouse genome, when using gatk to determine the copy number of chr19 using pipeline described in https://gatk.broadinstitute.org/hc/en-us/articles/360035531152--How-to-Call-rare-germline-copy-number-variants. The genotyped-segments outputted by PostprocessGermlineCNVCalls (default parameters with --autosomal-ref-copy-number 1, haploid cell) show whole chromosome duplication of chr19.

chr19 3200001 61300000 3076.53 540 2

I've checked the outputted denoised_copy_ratios.tsv file and found there was a clear cut between 27400001 and 27500001.

chr19 26900001 27000000 2.020929091459406
chr19 27000001 27100000 2.7166383354811
chr19 27100001 27200000 1.9484608134142096
chr19 27200001 27300000 1.7400198890920089
chr19 27300001 27400000 2.144735348146323
chr19 27400001 27500000 2.003273030134266 <------------
chr19 27500001 27600000 0.975046267767728 <------------
chr19 27600001 27700000 1.0879419961925338
chr19 27700001 27800000 0.8441336087343546
chr19 27800001 27900000 1.0025093782748349
chr19 27900001 28000000 0.9769124665764328

I've also checked the outputted mu_denoised_copy_ratio_t.tsv from GermlineCNVCaller (default parameters), it also shows the cut.

1.9484608134142096
1.7400198890920089
2.144735348146323
2.003273030134266 <------------
0.975046267767728 <------------
1.0879419961925338
0.8441336087343546
1.0025093782748349

However, in the log_q_c_tc.tsv file. Although a little bit smaller, value in COPY_NUMBER_2 is still the largest.

COPY_NUMBER_0 COPY_NUMBER_1 COPY_NUMBER_2 COPY_NUMBER_3 COPY_NUMBER_4 COPY_NUMBER_5
-22.08952932583426 -9.840231390864043 -0.00015831338758109936 -9.28860595034792 -11.406661531627021 -13.449075793524406
-22.08952932583426 -12.82969735638373 -0.0003148709940757888 -8.279217247759918 -9.906770240290832 -11.667322987231966
-22.08952932583426 -11.68657886797006 -0.0002128834781071065 -8.666335825294231 -10.487191052478696 -12.362959268138539 <------------
-22.08952932583426 -4.749089258681111 -0.008713068865176588 -11.115250385308611 -14.16405705250405 -16.766905053987283 <------------
-22.08952932583426 -5.391435367096327 -0.004584649717596756 -10.938173309564137 -13.929649290497585 -16.518027330281143
-22.08952932583426 -3.9163354815693925 -0.020126669802482955 -11.412369379149675 -14.602918767289168 -17.286594848265246

So, how should I change the parameters in PostprocessGermlineCNVCalls or GermlineCNVCaller to achieve

chr19 3200001 27500000 XXX DUP XXX 2
chr19 27500001 61300000 XXX NOR XXX 1