Alignpairedend: added alignment using shifting with best kmer similarity

(low level layer in C and Cython API)

python/obitools3/libalign/_solexapairend.pyx

@@ -3,6 +3,18 @@
 from cpython cimport array
 
 from .solexapairend import iterOnAligment
+from .shifted_ali cimport Ali_shifted
+
+from obitools3.dms.capi.obiview cimport Obiview_p, QUALITY_COLUMN, \
+                                        obi_set_qual_int_with_elt_idx_and_col_p_in_view, \
+                                        obi_set_str_with_elt_idx_and_col_p_in_view
+                                        
+from obitools3.dms.capi.obidmscolumn cimport OBIDMS_column_p
+
+from obitools3.dms.view.view cimport View
+from obitools3.dms.column.column cimport Column
+
+
 from math import log10
 
 
@@ -146,7 +158,7 @@ cdef class IterOnConsensus:
     seqBDeletion = property(get_seqBDeletion, None, None, "deletereverse's docstring")
 
 
-def buildConsensus(ali, seq):
+def buildConsensus(ali, seq, ref_tags=None):
     cdef list quality
     cdef char   aseq[1000]
     cdef int i=0
@@ -156,61 +168,82 @@ def buildConsensus(ali, seq):
     cdef double score
     cdef bytes sseq
     cdef double p
+    cdef OBIDMS_column_p col_p
+    cdef Obiview_p view_p
+    cdef View view
+    cdef Column column
     
     quality = []
+        
+    if type(ali) == Ali_shifted:
+        view = seq.view
+        view_p = view.pointer()
+        column = view[QUALITY_COLUMN]
+        col_p = column.pointer()
+        # doesn't work because uint8_t* are forced into bytes by cython (nothing read/kept beyond 0 values)
+        #obi_set_qual_int_with_elt_idx_and_col_p_in_view(view_p, col_p, seq.index, 0, ali.consensus_qual, ali.consensus_len)
+        seq.set(ref_tags.id+b"_CONS", ali.consensus_seq, quality=ali.consensus_qual, tags=ref_tags)
+        seq[b'ali_length'] = ali.consensus_len
+        seq[b'score_norm']=float(ali.score)/ali.consensus_len
+        seq[b'shift']=ali.shift
+    else:
+        if len(ali[0])>999:   # TODO why?
+            raise AssertionError,"Too long alignemnt"
     
-    if len(ali[0])>999:   # TODO why?
-        raise AssertionError,"Too long alignemnt"
-
-    ic=IterOnConsensus(ali)
-
-    for nuc,score in ic:
-        cnuc=nuc
-        quality.append(score)
-        aseq[i]=cnuc[0]
-        i+=1
-
-    aseq[i]=0
-
-    sseq=aseq
-
-    # Reconvert quality from array of probabilities to array of raw quality values
-    i=0
-    for p in quality:
-        quality[i] = min(42, round(-10*log10(p)))
-        i+=1
+        ic=IterOnConsensus(ali)
     
-    seq.set(ali[0].wrapped.id+b"_CONS", sseq, quality=quality, tags=ali[0].wrapped)
+        for nuc,score in ic:
+            cnuc=nuc
+            quality.append(score)
+            aseq[i]=cnuc[0]
+            i+=1
     
+        aseq[i]=0
+    
+        sseq=aseq
+    
+        # Reconvert quality from array of probabilities to array of raw quality values
+        i=0
+        for p in quality:
+            quality[i] = min(42, round(-10*log10(p)))
+            i+=1
+        
+        seq.set(ali[0].wrapped.id+b"_CONS", sseq, quality=quality, tags=ali[0].wrapped)
+        
+        if hasattr(ali, "counter"):
+            seq[b'alignement_id']=ali.counter
+        seq[b'seq_a_single']=ic.seqASingle
+        seq[b'seq_b_single']=ic.seqBSingle
+        seq[b'seq_ab_match']=ic.seqABMatch
+        seq[b'seq_a_mismatch']=ic.seqAMismatch
+        seq[b'seq_b_mismatch']=ic.seqBMismatch
+        seq[b'seq_a_insertion']=ic.seqAInsertion
+        seq[b'seq_b_insertion']=ic.seqBInsertion-ic.seqBSingle
+        seq[b'seq_a_deletion']=ic.seqADeletion
+        seq[b'seq_b_deletion']=ic.seqBDeletion
+        seq[b'ali_length']=len(seq)-ic.seqASingle-ic.seqBSingle
+        if seq[b'ali_length']>0:
+            seq[b'score_norm']=float(ali.score)/seq[b'ali_length']
+
     if hasattr(ali, "direction"):
-        seq[b'direction']=ali.direction
-    if hasattr(ali, "counter"):
-        seq[b'alignement_id']=ali.counter
-    seq[b'seq_a_single']=ic.seqASingle
-    seq[b'seq_b_single']=ic.seqBSingle
-    seq[b'seq_ab_match']=ic.seqABMatch
-    seq[b'seq_a_mismatch']=ic.seqAMismatch
-    seq[b'seq_b_mismatch']=ic.seqBMismatch
-    seq[b'seq_a_insertion']=ic.seqAInsertion
-    seq[b'seq_b_insertion']=ic.seqBInsertion-ic.seqBSingle
-    seq[b'seq_a_deletion']=ic.seqADeletion
-    seq[b'seq_b_deletion']=ic.seqBDeletion
+        seq[b'ali_direction']=ali.direction
     seq[b'score']=ali.score
-    seq[b'ali_length']=len(seq)-ic.seqASingle-ic.seqBSingle
-    if seq[b'ali_length']>0:
-        seq[b'score_norm']=float(ali.score)/seq[b'ali_length']
     seq[b'mode']=b'alignment'
+    
     return seq
 
 
-def buildJoinedSequence(ali, reverse, seq):
-    forward = ali[0].wrapped
+def buildJoinedSequence(ali, reverse, seq, forward=None):
+    if forward is not None:
+        forward = forward
+    else:
+        forward = ali[0].wrapped
     s = forward.seq + reverse.seq     
     quality = forward.quality
     quality.extend(reverse.quality)
     seq.set(forward.id +b"_PairedEnd", s, definition=forward.definition, quality=quality, tags=forward)
     seq[b"score"]=ali.score
-    seq[b"ali_dir"]=ali.direction
+    seq[b"ali_direction"]=ali.direction
     seq[b"mode"]=b"joined"
     seq[b"pairedend_limit"]=len(forward)    
     return seq