[modeller_usage] model generate error

[albert <>]

Hello:
    I am trying to use my own mutiple alignment to generate model, but it faild, it is said :

---------------log----------------------------------------
Protein specified in ALIGN_CODES(i) was not found
              in the alignment file; ALIGN_CODES(       4) =  protein


Here is parameter in my model-changeopt.py:
---------------model-changeopt.py----------------------
a = automodel(env, alnfile='man.ali', 
        knowns=('1A','2A','3A'), 
        sequence='protein',
        assess_methods=(assess.DOPE, assess.GA341))
a.starting_model =1
a.ending_model = 4
.......


Here is my alignment file, man.ali:
------------------------man.ali-------------------------------
>P1; protein
sequence:ttl:     : :     : ::: 0.00: 0.00
aaaaaaaaaaaaaabbbbbbbbbbbbbbcccccccccccccccccccc*
>P1;1A
structureX:1:    1    :A:+14 :A:::-1.00:-1.00
aaaaaaaaaaaaaa--------------------------------------------*
>P1;2A
structureX:2:    1    :A:+16  :A:::-1.00:-1.00
-----------------aabbbbbbbbbbbbbb--------------------------*
>P1;3A
structureX:3:    1    :A:+20 :A:::-1.00:-1.00
--------------------------------------cccccccccccccccccccc*

I have 1.pdb 2.pdb 3.pdb and 1_fit.pdb 2_fit.pdb 3_fit.pdb in the same directory.

could you please give me some advices please?
Thank you very much