[NEW] Added new "reldist" tool to measure the distribution of relative distances between two files

Makefile

@@ -54,6 +54,7 @@ SUBDIRS = $(SRC_DIR)/annotateBed \
 		  $(SRC_DIR)/pairToBed \
 		  $(SRC_DIR)/pairToPair \
 		  $(SRC_DIR)/randomBed \
+		  $(SRC_DIR)/reldist \
 		  $(SRC_DIR)/shuffleBed \
 		  $(SRC_DIR)/slopBed \
 		  $(SRC_DIR)/sortBed \

src/bedtools.cpp

@@ -63,6 +63,7 @@ int nuc_main(int argc, char* argv[]);//
 int pairtobed_main(int argc, char* argv[]);//
 int pairtopair_main(int argc, char* argv[]);//
 int random_main(int argc, char* argv[]); //
+int reldist_main(int argc, char* argv[]); //
 int shuffle_main(int argc, char* argv[]); //
 int slop_main(int argc, char* argv[]); //
 int sort_main(int argc, char* argv[]); //
@@ -126,6 +127,7 @@ int main(int argc, char *argv[])
 
     // statistics tools
     else if (sub_cmd == "jaccard")     return jaccard_main(argc-1, argv+1);
+    else if (sub_cmd == "reldist")     return reldist_main(argc-1, argv+1);
 
     // misc. tools
     else if (sub_cmd == "overlap")     return getoverlap_main(argc-1, argv+1);
@@ -230,8 +232,9 @@ int bedtools_help(void)
     cout  << "    tag           "  << "Tag BAM alignments based on overlaps with interval files.\n";
     
     cout  << endl;
-    cout  << "[ Statistics tools ]" << endl;
-    cout  << "    jaccard       "  << "Calculates the Jaccard statistic b/w two sets of intervals.\n";
+    cout  << "[ Statistical relationships ]" << endl;
+    cout  << "    jaccard       "  << "Calculate the Jaccard statistic b/w two sets of intervals.\n";
+    cout  << "    reldist       "  << "Calculate the distribution of relative distances b/w two files.\n";
 
     cout  << endl;
     cout  << "[ Miscellaneous tools ]" << endl;

test/reldist/test-reldist.sh

+BT=${BT-../../bin/bedtools}
+
+check()
+{
+	if diff $1 $2; then
+    	echo ok
+	else
+    	echo fail
+	fi
+}
+
+###########################################################
+#  Test a basic self intersection. The relative distances
+# should all be 0 in this case.
+############################################################
+echo "    reldist.t01...\c"
+echo \
+"reldist	count	total	fraction
+0.00	420889	420889	1.000" > exp
+$BT reldist -a ../../data/refseq.hg19.exons.bed.gz \
+            -b ../../data/refseq.hg19.exons.bed.gz > obs
+check obs exp
+rm obs exp
+
+###########################################################
+#  Test intervaks that are randomly distributed. 
+# The relative distances should equally represented .
+############################################################
+echo "    reldist.t02...\c"
+echo \
+"reldist	count	total	fraction
+0.00	1930	420584	0.005
+0.01	4937	420584	0.012
+0.02	6021	420584	0.014
+0.03	6774	420584	0.016
+0.04	7403	420584	0.018
+0.05	7152	420584	0.017
+0.06	7824	420584	0.019
+0.07	7818	420584	0.019
+0.08	7679	420584	0.018
+0.09	8275	420584	0.020
+0.10	7886	420584	0.019
+0.11	8308	420584	0.020
+0.12	8335	420584	0.020
+0.13	8532	420584	0.020
+0.14	8486	420584	0.020
+0.15	8335	420584	0.020
+0.16	8207	420584	0.020
+0.17	8674	420584	0.021
+0.18	8541	420584	0.020
+0.19	8795	420584	0.021
+0.20	8718	420584	0.021
+0.21	8731	420584	0.021
+0.22	8464	420584	0.020
+0.23	8627	420584	0.021
+0.24	8766	420584	0.021
+0.25	8927	420584	0.021
+0.26	8860	420584	0.021
+0.27	8876	420584	0.021
+0.28	9285	420584	0.022
+0.29	8882	420584	0.021
+0.30	9092	420584	0.022
+0.31	8882	420584	0.021
+0.32	8887	420584	0.021
+0.33	9296	420584	0.022
+0.34	8895	420584	0.021
+0.35	8964	420584	0.021
+0.36	9106	420584	0.022
+0.37	9377	420584	0.022
+0.38	9199	420584	0.022
+0.39	9240	420584	0.022
+0.40	9268	420584	0.022
+0.41	9443	420584	0.022
+0.42	9541	420584	0.023
+0.43	9287	420584	0.022
+0.44	8998	420584	0.021
+0.45	9238	420584	0.022
+0.46	8892	420584	0.021
+0.47	8859	420584	0.021
+0.48	9033	420584	0.021
+0.49	9028	420584	0.021
+0.50	11	420584	0.000" > exp
+$BT reldist -a ../../data/refseq.hg19.exons.bed.gz \
+            -b ../../data/aluY.hg19.bed.gz > obs
+check obs exp
+rm obs exp
+
+
+###########################################################
+#  Test intervaks that are consistently closer to one another
+# than expected.  The distances should be biased towards 0.=
+############################################################
+echo "    reldist.t03...\c"
+echo \
+"reldist	count	total	fraction
+0.00	156776	396736	0.395
+0.01	33579	396736	0.085
+0.02	17808	396736	0.045
+0.03	11826	396736	0.030
+0.04	8961	396736	0.023
+0.05	7302	396736	0.018
+0.06	5994	396736	0.015
+0.07	5357	396736	0.014
+0.08	5032	396736	0.013
+0.09	4415	396736	0.011
+0.10	4411	396736	0.011
+0.11	4062	396736	0.010
+0.12	3741	396736	0.009
+0.13	3760	396736	0.009
+0.14	3501	396736	0.009
+0.15	3394	396736	0.009
+0.16	3539	396736	0.009
+0.17	3442	396736	0.009
+0.18	3500	396736	0.009
+0.19	3072	396736	0.008
+0.20	3713	396736	0.009
+0.21	3249	396736	0.008
+0.22	3328	396736	0.008
+0.23	3271	396736	0.008
+0.24	3017	396736	0.008
+0.25	3756	396736	0.009
+0.26	3242	396736	0.008
+0.27	3057	396736	0.008
+0.28	3519	396736	0.009
+0.29	3222	396736	0.008
+0.30	3529	396736	0.009
+0.31	3284	396736	0.008
+0.32	3387	396736	0.009
+0.33	3231	396736	0.008
+0.34	3263	396736	0.008
+0.35	3421	396736	0.009
+0.36	3529	396736	0.009
+0.37	3376	396736	0.009
+0.38	3414	396736	0.009
+0.39	3183	396736	0.008
+0.40	3855	396736	0.010
+0.41	3402	396736	0.009
+0.42	3803	396736	0.010
+0.43	3458	396736	0.009
+0.44	3594	396736	0.009
+0.45	3654	396736	0.009
+0.46	3489	396736	0.009
+0.47	3394	396736	0.009
+0.48	3862	396736	0.010
+0.49	2957	396736	0.007
+0.50	805	396736	0.002" > exp
+$BT reldist -a ../../data/refseq.hg19.exons.bed.gz \
+            -b ../../data/29way_pi.conservation.hg19.bed.gz > obs
+check obs exp
+rm obs exp
\ No newline at end of file

test/test.sh

@@ -43,6 +43,9 @@ cd merge; sh test-merge.sh; cd ..
 echo " Testing bedtools multicov:"
 cd multicov; sh test-multicov.sh; cd ..
 
+echo " Testing bedtools reldist:"
+cd reldist; sh test-reldist.sh; cd ..
+
 echo " Testing bedtools slop:"
 cd slop; sh test-slop.sh; cd ..