As far as I understand, BER is the likelihood bit errors, say when a 0 becomes 1 or vice versa, so I would recommend to use the Corruption tab, where you can specify the percentage of packets to be corrupted. The software will flip a random bit in a packet to simulate a transmission error or a faulty network device.
Regarding the conversion, it is not possible to convert exactly. The BER of 10e-5 suggests that 1 bit in 10000 is damaged, which is 1 bit in every 1250 bytes. With the assumption that an average packet is 1500 bytes in size (MTU), this implies that 100% of packets are corrupted.
While you can set 100% packets to be corrupted, most likely no communication will be possible or it will be extremely slow with retransmissions. On the other hand, if your packets are smaller, say 125 bytes, the corruption rate should be much lower, 10%.
I guess it's best if we add a BER setting into the software directly, so the bit error rate could be expressed not only with percentage of packets, but also with BER specified directly. This may take a few days.