inconsistent scores for running on forward/reverse complement with non-uniform background probabilities

[st--]

Running MOODS manually on forward strand and reverse complement using -s seq_fw -m input.pfm -t 1 --lo-bg 0.23 0.19 0.24 0.34 and -s seq_rc -m input.pfm -t 1 --lo-bg 0.34 0.24 0.19 0.23 should result in the same hits and scores (except for which strand they are on). But the scores do not match:

This example is with the following sequence in forward (seq_fw):

>NS500683:269:H23L5AFXY:2:11302:11615:3251:1:N:0:GTCCGC
AAACGGTGCCGGGTTAGTTAAAACCTCTATGCGCCCTTTGGGAAATAATTAACGGCCGTT
TGTGGATAATTAAATTAGATTAATCATTTACTTTATGGTTAGTTGTTCTTCGTCATTGGG
ACGTTGAACTGATTCAGCTTTCTGTGCTTGTTTACCCCGTGCTGTTTGGG

and reverse-complement (seq_rc):

>NS500683:269:H23L5AFXY:2:11302:11615:3251:1:N:0:GTCCGC:rc
CCCAAACAGCACGGGGTAAACAAGCACAGAAAGCTGAATCAGTTCAACGTCCCAATGACG
AAGAACAACTAACCATAAAGTAAATGATTAATCTAATTTAATTATCCACAAACGGCCGTT
AATTATTTCCCAAAGGGCGCATAGAGGTTTTAACTAACCCGGCACCGTTT

and the following PFM (input.pfm):

70 52124 52124 4 13 52124 18696 52124 6 9 52124
24 4 0 1 122 1389 391 5 6 53 0
3 22 5 0 12 670 52124 135 0 0 20
52124 296 5 52124 52124 1241 11 10 52124 52124 215