wip

bwapy/libbwaaln.pyx

@@ -124,7 +124,6 @@ cdef class BwaAln:
     """The class to align reads with `bwa aln`."""
 
     cdef BwaIndex _index
-    cdef unsigned char* _pacseq
 
     def __init__(self, prefix: str | Path | None = None, index: BwaIndex | None = None):
         if prefix is not None:

bwapy/libbwamem.pxd

@@ -1,8 +1,11 @@
 # cython: language_level=3
 
-from libc.stdint cimport uint8_t, int64_t, int32_t, uint64_t, int8_t
+from libc.stdint cimport uint8_t, int64_t, int32_t, uint64_t, int8_t, uint32_t
 from libc.stdio cimport FILE
 
+cdef extern from "limits.h":
+    cdef int INT_MAX
+
 cdef extern from "bwt.h":
     ctypedef struct bwt_t:
         int sa_intv
@@ -19,6 +22,13 @@ cdef extern from "bntseq.h":
         bntann1_t *anns
         FILE * fp_pac
 
+    unsigned char nst_nt4_table[256]
+
+cdef extern from "kstring.h":
+    ctypedef struct kstring_t:
+        size_t l, m
+        char *s
+
 
 cdef extern from "bwamem.h":
     int MEM_F_PE        
@@ -70,9 +80,32 @@ cdef extern from "bwamem.h":
         int max_matesw         # perform maximally max_matesw rounds of mate-SW for each end
         int max_XA_hits, max_XA_hits_alt # if there are max_hits or fewer, output them all
         int8_t mat[25]         # scoring matrix mat[0] == 0 if unset
+    ctypedef struct mem_alnreg_t:
+        int score  # best local SW score
+        int secondary  # index of the parent hit shadowing the current hit; <0 if primary
+        int n_comp  # number of sub-alignments chained together
+        int is_alt
 
     ctypedef struct mem_alnreg_v:
-        pass
+        size_t n, m
+        mem_alnreg_t *a
+
+    ctypedef struct mem_aln_t:
+        int flag # extra flag
+        uint32_t is_rev  # is_rev: whether on the reverse strand;
+        uint32_t is_alt
+        uint32_t mapq   # mapq: mapping quality;
+        uint32_t NM  # NM: edit distance
+        char *XA
+        int score, sub, alt_sc;
+
     mem_opt_t *mem_opt_init()
     mem_alnreg_v mem_align1(const mem_opt_t *opt, const bwt_t *bwt, const bntseq_t *bns, const uint8_t *pac, int l_seq, const char *seq)
-    void bwa_fill_scmat(int a, int b, int8_t mat[25])
+    void bwa_fill_scmat(int a, int b, int8_t mat[25])
+    mem_aln_t mem_reg2aln(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, int l_seq, const char *seq, const mem_alnreg_t *ar)
+
+# from bwamem.c
+cdef extern void mem_reorder_primary5(int T, mem_alnreg_v *a)
+
+# from bwamem_extra.c
+cdef extern char **mem_gen_alt(const mem_opt_t *opt, const bntseq_t *bns, const uint8_t *pac, mem_alnreg_v *a, int l_query, const char *query);

bwapy/libbwamem.pyi

@@ -1,12 +1,24 @@
+from pathlib import Path
+from typing import List
 
+from pysam import AlignedSegment
+from pysam import FastxRecord
 
+from bwapy.libbwaindex import BwaIndex
 
 class BwaMemOptions:
     _ignore_alt: bool
     def __init__(self) -> None: ...
 
-
 class BwMemOptionsBuilder:
     _options: BwaMemOptions
+    _options0: BwaMemOptions
     def __init__(self, options: BwaMemOptions | None = None) -> None: ...
-    def build(self) -> BwaMemOptions: ...
+    def build(self) -> BwaMemOptions: ...
+
+class BwaMem:
+    _index: BwaIndex
+    def __init__(self, prefix: str | Path | None = None, index: BwaIndex | None = None) -> None: ...
+    def align(
+        self, opt: BwaMemOptions, queries: List[FastxRecord]
+    ) -> List[List[AlignedSegment]]: ...