add end of sequences in as points in our array

2 files changed, 21 insertions, 1 deletions

diff --git a/src/score_mers.py b/src/score_mers.py
index 6f0c6db..5c1a194 100755
--- a/src/score_mers.py
+++ b/src/score_mers.py
@@ -15,6 +15,8 @@ import pdb
 fg_mers = {}
 bg_mers = {}
 
+seq_ends = []
+
 if(len(sys.argv) == 5):
 	selectivity_fn =  sys.argv[1]
 	fg_fasta_fn    =  sys.argv[2]
@@ -181,8 +183,13 @@ def score(combination):
 	for mer in combination:
 		fg_pts = fg_pts + fg_mers[mer].pts
 
+	fg_pts = fg_pts + seq_ends 
+
 	fg_pts.sort()
 
+	if fg_pts[0] is not 0:
+		fg_pts = [0] + fg_pts
+
 	# fg distances
 	fg_dist = np.diff(fg_pts)
 
@@ -221,6 +228,18 @@ def score(combination):
 
 	return [combination, score, fg_mean_dist, fg_std_dist, bg_ratio] 
 
+def load_end_points(fn):
+
+	end_points = []
+
+	cmd = "sequence_end_points < " + fn
+
+	points_fh = Popen(cmd, stdout=PIPE, shell=True)
+	for line in points_fh.stdout:
+		end_points.append(int(line))
+	
+	return end_points
+
 def load_heterodimer_dic(selected_mers):
 	'''
 	Generate a heterodimer dict which contains every possible combination of
@@ -265,6 +284,8 @@ def main():
 	selected_mers = [row[0] for row in selected]
 	# print selected_mers
 
+	print "Populating sequence end points"
+	seq_ends = load_end_points(fg_fasta_fn)
 	print "Populating foreground locations"
 	map(pop_fg, selected_mers)
 
diff --git a/src/sequence_length.c b/src/sequence_end_points.c
index 3df7175..dc03dd9 100644
--- a/src/sequence_length.c
+++ b/src/sequence_end_points.c
@@ -24,7 +24,6 @@ int main() {
 				for(i = 0; i < len; i++) {
 					if(buffer[i] == '>') {
 						printf("%llu\n", seq_length);
-						seq_length = 0;
 						header = true;
 						continue;
 					}