I am a bit curious of your request. There is some background that you may find of interest, regarding the nature of symbol sequences. Please see this discussion -

https://www.researchgate.net/post/Not_a_question_but_an_invitation_to_discussion-Shannon_Chaitin_and_the_genome

Generally, one expects that a primer will be of length close to the expected shustring length of a nucleotide sequence, as suggested in the above discussion. For human, these should be something like 15 to 20 nucleotides; this will typically give selection of just a few loci.

If you know the banana sequence (has it been fully sequenced ?), and you know the locus of interest, you should be able quickly to determine an highly selective primer.