some analysis about barrals with sandwich's HMM

g1rin.1: 13B.fa 121.5 5.1e-35 1
looks more like a sandwich to me then barral.

g1ghw.1: 1F.fa 121.2 6.2e-35 1
looks like barral, but has interlock.

d1eu3a1: 20B.fa -46.2 1.8e+02 1
a barral looks like sandwich.
but I agree it's a barral, since I didn't saw a sandwich in this angle.

d3vub__: 20B.fa -49.6 1.8e+02 1
two barral share one wall, like "8" shape

d1jb3a_: 20B.fa -42.4 1.8e+02 1
barral
1 2 3 5 4

4 1 2'
5 3 2

d3chbd_: 20B.fa -44.5 1.8e+02 1
motif 20B only has one structure which doesn't looks like a regular sandwich.
d3chbd_ looks like a distorted barral, which looks like a distorted sandwich also.

d1jzua_: 20B.fa -36.6 1.8e+02 1
jelly roll, probably because of the strands length, it almost form a sandwich like structure.

3V
a motif doesn't have interlock.

d1qb5d_: 3V.fa -44.2 1.8e+02 1
there's one sheet which looks like sandwich's sheet.

d1g7sa1: 3V.fa -45.3 1.8e+02 1
probably because the additional sheets in barral

some of these structures have alpha helix, so our HMM probably can't recognize it perfectly.

d1ewna_: 2G.fa -22.0 1.8e+02 1
have lots of alpha helix

d1ja1a1: 2G.fa -10.2 1.8e+02 1
have lots of alpha helix in this domain

d1f20a1: 2G.fa -13.9 1.8e+02 1
have lots of alpha helix in this domain

d1i4ua_: 2G.fa -29.0 1.8e+02 1

the training set we train HMM has already been modified before grouping, but the sequences we train it doesn't.