added test

Author: lmdu

.gitignore

@@ -40,3 +40,6 @@ chr20.fa
 chr20.fa.fxi
 hg19.fa
 stria.cp38-win_amd64.pyd
+Homo_sapiens.GRCh38.dna.toplevel.fa.gz.fxi
+Homo_sapiens.GRCh38.dna.toplevel.fa.gz
+stria.cpython-310-x86_64-linux-gnu.so

benchmark.py

@@ -0,0 +1,17 @@
+import time
+import stria
+import pyfastx
+gfile = 'Homo_sapiens.GRCh38.dna.toplevel.fa.gz'
+
+for name, seq in pyfastx.Fastx(gfile):
+	break
+
+finder = stria.SSRMiner(name, seq)
+
+start = time.time()
+ssrs = finder.as_list()
+print(time.time() - start)
+
+start = time.time()
+ssrs = finder.as_test()
+print(time.time() - start)

src/ssr.c

@@ -217,8 +217,58 @@ static PyObject* stria_ssrminer_as_list(stria_SSRMiner *self) {
 	return ssrs;
 }
 
+static PyObject* stria_ssrminer_as_test(stria_SSRMiner *self) {
+	PyObject *ssrs = PyList_New(0);
+	PyObject *tmp;
+	Py_ssize_t current_start;
+	Py_ssize_t ssr_end;
+	Py_ssize_t boundary;
+	int replen;
+	int repeats;
+	int length;
+	//char *motif = (char *)malloc(7);
+	char motif[7];
+
+	for (Py_ssize_t i = 0; i < self->size; ++i) {
+		if (self->seq[i] == 78) {
+			continue;
+		}
+
+		current_start = i;
+		for (int j = 1; j < 7; ++j) {
+			boundary = self->size - j;
+
+			while ((i < boundary) && (self->seq[i] == self->seq[i+j])) {
+				++i;
+			}
+
+			replen = i + j - current_start;
+
+			if (replen >= self->min_lens[j]) {
+				memcpy(motif, self->seq+current_start, j);
+				motif[j] = '\0';
+				repeats = replen / j;
+				length = repeats * j;
+				ssr_end = current_start+length;
+				tmp = Py_BuildValue("Onnsiii", self->seqname, current_start+1, ssr_end, motif, j, repeats, length);
+				PyList_Append(ssrs, tmp);
+				Py_DECREF(tmp);
+
+				i = ssr_end;
+				break;
+			} else {
+				i = current_start;
+			}
+		}
+	}
+
+	//free(motif);
+	return ssrs;
+}
+
 static PyMethodDef stria_ssrminer_methods[] = {
 	{"as_list", (PyCFunction)stria_ssrminer_as_list, METH_NOARGS, NULL},
+	{"as_test", (PyCFunction)stria_ssrminer_as_test, METH_NOARGS, NULL},
 	{"reset_min_repeats", (PyCFunction)stria_ssrminer_reset_min_repeats, METH_VARARGS|METH_KEYWORDS, NULL},
 	{NULL, NULL, 0, NULL}
 };

test.py

@@ -1,10 +1,63 @@
+import sys
 import time
-import stria
 import pyfastx
 
-start = time.time()
-for name, seq, _ in pyfastx.Fastx('chr1.fa'):
-	ssrs = stria.test(name, seq)
-	#ssrs = stria.SSRMiner(name,seq).as_list()
-print(time.time()-start)
-print(len(ssrs))
+start_time = time.time()
+mins = [0, 12, 14, 15, 16, 20, 24]
+
+for name, seq in pyfastx.Fastx(sys.argv[1]):
+	pass
+
+size = len(seq)
+count = 0
+i = 0
+jump = 0
+
+while i < size:
+	if seq[i] == 'N':
+		i += 1
+		continue
+
+	start = i
+	run = [0, 0, 0, 0, 0, 0, 0]
+
+	for j in range(1, 7):
+		'''
+		if j > 1:
+			if run[1] >= j:
+				jump += 1
+				continue
+
+			if j == 4 and run[2] >= j:
+				jump += 1
+				continue
+
+			elif j == 6 and (run[2] >= j or run[3] >= j):
+				jump += 1
+				continue
+		'''
+
+		b = size - j
+
+		while i < b and seq[i] == seq[i+j]:
+			i += 1
+
+		run[j] = i + j - start
+
+		if run[j] >= mins[j]:
+			count += 1
+			repeat = run[j]//j
+			i = start + repeat*j
+			run = [0, 0, 0, 0, 0, 0, 0]
+			break
+		else:
+			i = start
+
+	i += 1
+
+end_time = time.time()
+
+print(end_time-start_time)
+print(count)
+print(jump)
+print(size)