Pepspec - sequence sampling

Hello,

I am using Pepspec to optimize a 10mers peptide sequence to better binding to a membrane protein. In the design, I kept the first residue constant, while mutating the rest by all 20 amino acids. The flags file is the following:

-in:file:s ppk_complex_0001.pdb

-pepspec::pep_chain B
-pepspec::pep_anchor 1
-pepspec::n_append 9
-pepspec::n_peptides 10000

-ignore_unrecognized_res
-ex1
-ex2
-extrachi_cutoff 0

-score::weights ref2015.wts
-pepspec::soft_wts soft_rep.wts

-pepspec::interface_cutoff
-pepspec::clash_cutoff 5.0

-pepspec::n_build_loop 100
-pepspec::diversify_pep_seqs
-pepspec::diversify_lvl 10
-pepspec:remove_input_bb
-pepspec:gen_pep_bb_sequential

-pepspec::save_low_pdbs

-o pepspec_complex

-out:file:design_contrast redesign_complex
-out:file:silent pepspec_complex_redesign.silent         
-out:file:silent_struct_type binary
-out:file:fullatom
-out:file:scorefile pepspec_complex_redesign

The run generated 990668 non-redundant sequences and saved 10000 pdbs. Then I used WebLogo to evaluate the quality of the sampling (see attached picture). As can be seen, there is an over-enrichment of Ser and Gly along the sequence, which looks weird to me. Is this an artifact of the sampling? How can I improve the sampling?

Thanks in advance,

Yasser