correct enc_r and encp_r's POD

fortune9 · May 3, 2015 · a31339f · a31339f
1 parent 43a4822
commit a31339f
Show file tree

Hide file tree

Showing 3 changed files with 100 additions and 102 deletions.
diff --git a/lib/Bio/CUA/CUB/Calculator.pm b/lib/Bio/CUA/CUB/Calculator.pm
@@ -377,19 +377,18 @@ sub enc
 
  Title   : enc_r
  Usage   : $encValue = $self->enc_r($seq,[$minTotal]);
- Function: similar to the method L</enc>, except that calculating the
- missing F values using a different way, relying on the relationship
- of homozygosity and degree of degeneracy.
+ Function: similar to the method L</enc>, except that missing F values
+ are estimated in a different way.
  Returns : a number, or undef if failed
  Args    : for sequence see L</"sequence input">.
  Optional argument I<minTotal> specifies minimal count 
  for an amino acid; if observed count is smaller than this count, this
  amino acid's F will not be calculated but inferred. Deafult is 5.
 
- Note: its difference from method L</enc> is to estimate missing Fr
- using equal_ratio algorithm, that is the estimate Fx satisfying the
- following equation: (1/Fx)/x=[sum(1/Fr) over r]/[sum(r) over r] where
- x and r marks groups with unknown and known F values, respectively.
+ Note: for missing Fx of degeneracy class 'x', we first estimated the
+ ratio (1/Fx-1)/(x-1) by averaging the ratios of other degeneracy
+ classes with known F values. Then Fx is obtained by solving the simple
+ equation.
 
 =cut
 
@@ -431,9 +430,8 @@ sub encp
  Title   : encp_r
  Usage   : $encpValue =
  $self->encp_r($seq,[$minTotal,[$A,$T,$C,$G]]);
- Function: similar to the method L</encp>, except that calculating the
- missing F values using a different way, considering the
- redundancy degree of each class.
+ Function: similar to the method L</encp>, except that missing F values
+ are estimated using a different way.
  Returns : a number, or undef if failed
  Args    : for sequence see L</"sequence input">.
  
@@ -446,10 +444,10 @@ sub encp
  default one provided when calling the method L</new>. If stil
  unavailable, error occurs.
 
- Note: its difference from method L</encp> is to estimate missing Fr
- using equal_ratio algorithm, that is the estimate Fx satisfying the
- following equation (1/Fx)/x=[sum(1/Fr) over r]/[sum(r) over r] where
- x and r label unknown and known degeneracy groups, respectively.
+ Note: for missing Fx of degeneracy class 'x', we first estimated the
+ ratio (1/Fx-1)/(x-1) by averaging the ratios of other degeneracy
+ classes with known F values. Then Fx is obtained by solving the simple
+ equation.
 
 =cut
 

diff --git a/t/cai.out b/t/cai.out
@@ -1,59 +1,59 @@
-CCG	0.878306878306878
-ACA	0.469767441860465
+CGA	0.364864864864865
+CTC	0.386058981233244
+TAT	0.537634408602151
+CCC	1
+TGC	1
 GTT	0.330316742081448
-AGG	0.310810810810811
+TAC	1
+GTG	1
+ATT	0.414225941422594
+AGA	0.243243243243243
+CTA	0.190348525469169
+AAT	0.604477611940298
 TGT	0.23
-CGT	0.371621621621622
-GCA	0.274336283185841
+TTA	0.0857908847184987
 AGC	1
+GCA	0.274336283185841
+CTG	1
+GGT	0.335616438356164
+CTT	0.142091152815013
+AGT	0.450184501845018
 ACT	0.288372093023256
 GAT	0.892857142857143
-ATC	1
-TGC	1
 CAC	1
-CCA	0.587301587301587
-TCA	0.254612546125461
-CTC	0.386058981233244
+TTC	1
 AAA	0.398576512455516
-ACG	0.744186046511628
 CCT	0.407407407407407
-CTT	0.142091152815013
 TTG	0.329758713136729
-AAT	0.604477611940298
-ATA	0.284518828451883
+CGC	1
 GAG	1
-ATT	0.414225941422594
-CCC	1
-CGA	0.364864864864865
-GTA	0.194570135746606
+AGG	0.310810810810811
+CGG	0.540540540540541
 TCC	0.845018450184502
-TAT	0.537634408602151
-GAC	1
-TCT	0.191881918819188
-CGC	1
-AAC	1
+ACG	0.744186046511628
+CCG	0.878306878306878
+GGA	0.537671232876712
 GAA	0.394927536231884
-GCT	0.309734513274336
-CAT	0.537688442211055
+GGC	1
 CAA	0.380352644836272
+GCC	1
+AAG	1
+CGT	0.371621621621622
+ACA	0.469767441860465
+ACC	1
+TCG	0.712177121771218
+GAC	1
 CAG	1
-CTA	0.190348525469169
 GGG	0.174657534246575
-ACC	1
 GCG	0.47787610619469
-TCG	0.712177121771218
-TAC	1
-AGA	0.243243243243243
-TTA	0.0857908847184987
-GGT	0.335616438356164
-CTG	1
-GTC	0.565610859728507
-TTC	1
-GGC	1
+GCT	0.309734513274336
+GTA	0.194570135746606
 TTT	0.450261780104712
-GTG	1
-GCC	1
-CGG	0.540540540540541
-AGT	0.450184501845018
-GGA	0.537671232876712
-AAG	1
+GTC	0.565610859728507
+ATC	1
+ATA	0.284518828451883
+TCA	0.254612546125461
+CCA	0.587301587301587
+CAT	0.537688442211055
+TCT	0.191881918819188
+AAC	1
diff --git a/t/tai.out b/t/tai.out
@@ -1,60 +1,60 @@
-ATC	0.309751434034417
-ACT	0.430210325047801
-GAT	0.394837476099426
+CTA	0.0956261950286807
+GTG	0.365200764818356
+ATT	0.430210325047801
+TAC	0.430210325047801
+AGA	0.1434034416826
+CCC	0.240917782026769
 TGC	0.334608030592734
-AGC	0.286806883365201
-CAC	0.239005736137667
-TGG	0.397705544933078
-TCA	0.095640535372849
-CCA	0.239039196940727
-CTC	0.17208413001912
-ACG	0.235181644359465
-AAA	0.286806883365201
-TTG	0.252390057361377
-CTT	0.239005736137667
-CCT	0.334608030592734
-ATA	0.0956453154875717
-AAT	0.338432122370937
-CCG	0.315487571701721
-AGG	0.189292543021033
 GTT	0.286806883365201
-ACA	0.286849904397706
-TGT	0.197418738049713
+CGA	0.478059273422562
+CTC	0.17208413001912
+TAT	0.253824091778203
 GCA	0.09565965583174
-CGT	0.478011472275335
-GCG	0.173996175908222
-TCG	0.221797323135755
-AGA	0.1434034416826
-TAC	0.430210325047801
 CTG	0.413001912045889
 TTA	0.191204588910134
-GGT	0.394837476099426
+AGC	0.286806883365201
+AAT	0.338432122370937
+TGT	0.197418738049713
 TTC	0.382409177820268
-GGC	0.669216061185469
-GTC	0.206500956022945
-GCC	0.413001912045889
-GTG	0.365200764818356
-TTT	0.225621414913958
+AAA	0.286806883365201
+GAT	0.394837476099426
+CAC	0.239005736137667
+GGT	0.394837476099426
+ACT	0.430210325047801
+CTT	0.239005736137667
 AGT	0.169216061185469
-CGG	0.152963671128107
-AAG	0.713193116634799
+CGC	0.344168260038241
+TTG	0.252390057361377
+CCT	0.334608030592734
+CAA	0.191204588910134
+GCC	0.413001912045889
+TCC	0.275334608030593
+ACG	0.235181644359465
+GGC	0.669216061185469
+CCG	0.315487571701721
+GAA	0.286806883365201
 GGA	0.286806883365201
-CCC	0.240917782026769
-ATT	0.430210325047801
+CGG	0.152963671128107
+AGG	0.189292543021033
 GAG	1
+ACC	0.309751434034417
+ACA	0.286849904397706
+TGG	0.397705544933078
+AAG	0.713193116634799
+CGT	0.478011472275335
+GCT	0.573613766730402
 GTA	0.0956309751434034
-CGA	0.478059273422562
-TAT	0.253824091778203
+GCG	0.173996175908222
 GAC	0.669216061185469
-TCC	0.275334608030593
-GAA	0.286806883365201
+CAG	0.443594646271511
+GGG	0.0917782026768642
+TCG	0.221797323135755
 AAC	0.573613766730402
+CCA	0.239039196940727
+TCA	0.095640535372849
 TCT	0.382409177820268
-CGC	0.344168260038241
-GCT	0.573613766730402
-CAG	0.443594646271511
-CAA	0.191204588910134
 CAT	0.141013384321224
-CTA	0.0956261950286807
-GGG	0.0917782026768642
-ACC	0.309751434034417
+TTT	0.225621414913958
+ATA	0.0956453154875717
+ATC	0.309751434034417
+GTC	0.206500956022945