added a recoding function for alignments. NT and AA alphabets can now…

… be reduced down to fewer characters

docs/usage/index.rst

@@ -138,6 +138,122 @@ Options: |br|
 
 |
 
+Alignment recoding 
+##################
+Function names: alignment_recoding; aln_recoding; recode |br|
+Command line interface: pk_alignment_recoding; pk_aln_recoding; bk_recode
+
+Recode alignments using reduced character states.
+
+Alignments can be recoded using established or
+custom recoding schemes. Recoding schemes are
+specified using the -c/--code argument. Custom
+recoding schemes can be used and should be formatted
+as a two column file wherein the first column is the
+recoded character and the second column is the character
+in the alignment.
+
+.. code-block:: shell
+
+	phykit alignment_recoding <fasta> [-c/--code <code>]
+
+Codes for which recoding scheme to use: |br|
+**RY-nucleotide** |br|
+R = purines (i.e., A and G) |br|
+Y = pyrimidines (i.e., T and C) |br|
+
+**SandR-6** |br|
+0 = A, P, S, and T |br|
+1 = D, E, N, and G |br|
+2 = Q, K, and R |br|
+3 = M, I, V, and L |br|
+4 = W and C |br|
+5 = F, Y, and H |br|
+
+**KGB-6** |br|
+0 = A, G, P, and S |br|
+1 = D, E, N, Q, H, K, R, and T |br|
+2 = M, I, and L |br|
+3 = W |br|
+4 = F and Y |br|
+5 = C and V |br|
+
+**Dayhoff-6** |br|
+0 = A, G, P, S, and T |br|
+1 = D, E, N, and Q |br|
+2 = H, K, and R |br|
+3 = I, L, M, and V |br|
+4 = F, W, and Y |br|
+5 = C |br|
+
+**Dayhoff-9** |br|
+0 = D, E, H, N, and Q |br|
+1 = I, L, M, and V |br|
+2 = F and Y |br|
+3 = A, S, and T |br|
+4 = K and R |br|
+5 = G |br|
+6 = P |br|
+7 = C |br|
+8 = W |br|
+
+**Dayhoff-12** |br|
+0 = D, E, and Q |br|
+1 = M, L, I, and V |br|
+2 = F and Y |br|
+3 = K, H, and R |br|
+4 = G |br|
+5 = A |br|
+6 = P |br|
+7 = S |br|
+8 = T |br|
+9 = N |br|
+A = W |br|
+B = C |br|
+
+**Dayhoff-15** |br|
+0 = D, E, and Q |br|
+1 = M and L |br|
+2 = I and V |br|
+3 = F and Y |br|
+4 = G |br|
+5 = A |br|
+6 = P |br|
+7 = S |br|
+8 = T |br|
+9 = N |br|
+A = K |br|
+B = H |br|
+C = R |br|
+D = W |br|
+E = C |br|
+
+**Dayhoff-18** |br|
+0 = F and Y |br|
+1 = M and L |br|
+2 = I |br|
+3 = V |br|
+4 = G |br|
+5 = A |br|
+6 = P |br|
+7 = S |br|
+8 = T |br|
+9 = D |br|
+A = E |br|
+B = Q |br|
+C = N |br|
+D = K |br|
+E = H |br|
+F = R |br|
+G = W |br|
+H = C |br|
+
+Options: |br|
+*<alignment>*: first argument after function name should be an alignment file |br|
+*-c/\-\-code*: argument to specify the recoding scheme to use 
+
+|
+
 Column score
 ############
 

phykit/phykit.py

@@ -15,6 +15,7 @@
 from .services.alignment import (
     AlignmentLength,
     AlignmentLengthNoGaps,
+    AlignmentRecoding,
     ColumnScore,
     CreateConcatenationMatrix,
     DNAThreader,
@@ -144,6 +145,8 @@ def __init__(self):
                     - calculates alignment length
                 alignment_length_no_gaps (alias: aln_len_no_gaps; alng)
                     - calculates alignment length after removing sites with gaps
+                alignment_recoding (alias: aln_recoding, recode)
+                    - recode alignments using reduced character schemes
                 column_score (alias: cs)
                     - calculate column score between a reference and query alignment
                 create_concatenation_matrix (alias: create_concat; cc)
@@ -265,6 +268,8 @@ def run_alias(self, command, argv):
             return self.alignment_length(argv)
         elif command in ["aln_len_no_gaps", "alng"]:
             return self.alignment_length_no_gaps(argv)
+        elif command in ["aln_recoding", "recode"]:
+            return self.alignment_recoding(argv)
         elif command in "cs":
             return self.column_score(argv)
         elif command in ["get_entry", "ge"]:
@@ -451,6 +456,142 @@ def alignment_length_no_gaps(argv):
         args = parser.parse_args(argv)
         AlignmentLengthNoGaps(args).run()
 
+    @staticmethod
+    def alignment_recoding(argv):
+        parser = ArgumentParser(
+            add_help=True,
+            usage=SUPPRESS,
+            formatter_class=RawDescriptionHelpFormatter,
+            description=textwrap.dedent(
+                f"""\
+                {help_header}
+
+                Recode alignments using reduced character states.
+
+                Alignments can be recoded using established or
+                custom recoding schemes. Recoding schemes are
+                specified using the -c/--code argument. Custom
+                recoding schemes can be used and should be formatted
+                as a two column file wherein the first column is the
+                recoded character and the second column is the character
+                in the alignment.
+                
+                Aliases:
+                  alignment_recoding, aln_recoding, recode
+                Command line interfaces: 
+                  bk_alignment_recoding, bk_aln_recoding, bk_recode
+
+                Usage:
+                phykit alignment_recoding <fasta> -c/--code <code>
+
+                Options
+                =====================================================
+                <fasta>                     first argument after 
+                                            function name should be
+                                            a fasta file
+
+                -c/--code                   recoding scheme to use
+
+                Codes for which recoding scheme to use
+                =====================================================
+                RY-nucleotide
+                    R = purines (i.e., A and G) 
+                    Y = pyrimidines (i.e., T and C)
+                
+                SandR-6
+                    0 = A, P, S, and T
+                    1 = D, E, N, and G
+                    2 = Q, K, and R
+                    3 = M, I, V, and L
+                    4 = W and C
+                    5 = F, Y, and H
+
+                KGB-6
+                    0 = A, G, P, and S
+                    1 = D, E, N, Q, H, K, R, and T
+                    2 = M, I, and L
+                    3 = W
+                    4 = F and Y
+                    5 = C and V
+
+                Dayhoff-6
+                    0 = A, G, P, S, and T
+                    1 = D, E, N, and Q
+                    2 = H, K, and R
+                    3 = I, L, M, and V
+                    4 = F, W, and Y
+                    5 = C
+
+                Dayhoff-9
+                    0 = D, E, H, N, and Q
+                    1 = I, L, M, and V
+                    2 = F and Y
+                    3 = A, S, and T
+                    4 = K and R
+                    5 = G
+                    6 = P
+                    7 = C
+                    8 = W
+
+                Dayhoff-12
+                    0 = D, E, and Q
+                    1 = M, L, I, and V
+                    2 = F and Y
+                    3 = K, H, and R
+                    4 = G
+                    5 = A
+                    6 = P
+                    7 = S
+                    8 = T
+                    9 = N
+                    A = W
+                    B = C
+                
+                Dayhoff-15
+                    0 = D, E, and Q
+                    1 = M and L
+                    2 = I and V
+                    3 = F and Y
+                    4 = G
+                    5 = A
+                    6 = P
+                    7 = S
+                    8 = T
+                    9 = N
+                    A = K
+                    B = H
+                    C = R
+                    D = W
+                    E = C
+                
+                Dayhoff-18
+                    0 = F and Y
+                    1 = M and L
+                    2 = I
+                    3 = V
+                    4 = G
+                    5 = A
+                    6 = P
+                    7 = S
+                    8 = T
+                    9 = D
+                    A = E
+                    B = Q
+                    C = N
+                    D = K
+                    E = H
+                    F = R
+                    G = W
+                    H = C
+                """  # noqa
+            ),
+        )
+
+        parser.add_argument("alignment", type=str, help=SUPPRESS)
+        parser.add_argument("-c", "--code", type=str, help=SUPPRESS)
+        args = parser.parse_args(argv)
+        AlignmentRecoding(args).run()
+
     @staticmethod
     def column_score(argv):
         parser = ArgumentParser(

phykit/recoding_tables/Dayhoff-12.txt

@@ -0,0 +1,20 @@
+0       D
+0       E
+0       Q
+1       M
+1       L
+1       I
+1       V
+2       F
+2       Y
+3       K
+3       H
+3       R
+4       G
+5       A
+6       P
+7       S
+8       T
+9       N
+A       W
+B       C

phykit/recoding_tables/Dayhoff-15.txt

@@ -0,0 +1,20 @@
+0       D
+0       E
+0       Q
+1       M
+1       L
+2       I
+2       V
+3       F
+3       Y
+4       G
+5       A
+6       P
+7       S
+8       T
+9       N
+A       K
+B       H
+C       R
+D       W
+E       C

phykit/recoding_tables/Dayhoff-18.txt

@@ -0,0 +1,20 @@
+0       F
+0       Y
+1       M
+1       L
+2       I
+3       V
+4       G
+5       A
+6       P
+7       S
+8       T
+9       D
+A       E
+B       Q
+C       N
+D       K
+E       H
+F       R
+G       W
+H       C

phykit/recoding_tables/Dayhoff-6.txt

@@ -0,0 +1,20 @@
+0   A
+0   G
+0   P
+0   S
+0   T
+1   D
+1   E
+1   N
+1   Q
+2   H
+2   K
+2   R
+3   I
+3   L
+3   M
+3   V
+4   F
+4   W
+4   Y
+5   C

phykit/recoding_tables/Dayhoff-9.txt

@@ -0,0 +1,20 @@
+0       D
+0       E
+0       H
+0       N
+0       Q
+1       I
+1       L
+1       M
+1       V
+2       F
+2       Y
+3       A
+3       S
+3       T
+4       K
+4       R
+5       G
+6       P
+7       C
+8       W