diff --git a/resources/curated/somatic_hypermutation_locations_GRCh37.txt b/resources/curated/somatic_hypermutation_locations_GRCh37.txt
index b83ac93f..c4614940 100644
--- a/resources/curated/somatic_hypermutation_locations_GRCh37.txt
+++ b/resources/curated/somatic_hypermutation_locations_GRCh37.txt
@@ -1,89 +1,130 @@
-chr_name	hg19_start	hg19_end	gene	region
-chr1	203274698	203275778	BTG2	intronic
-chr1	226864857	226873452	ITPKB	intronic
-chr1	226920563	226927885	ITPKB	TSS
-chr1	226921088	226927982	ITPKB	intron-1
-chr10	127578912	127591133	FANK1	TSS
-chr11	35156769	35164248	CD44	TSS
-chr11	60223385	60225310	MS4A1	TSS
-chr11	65190369	65192380	NEAT1	ncRNA
-chr11	65265237	65268359	MALAT1	ncRNA
-chr11	102188170	102190077	BIRC3	TSS
-chr11	111248078	111250224	POU2AF1	TSS
-chr11	128339774	128345731	ETS1	introns
-chr11	128388492	128394163	ETS1	TSS-2
-chr12	11796001	11812968	ETV6	TSS
-chr12	25204045	25213569	LRMP	TSS
-chr12	92537999	92539598	BTG1	TSS
-chr12	113492311	113497546	DTX1	TSS
-chr12	122456912	122464036	BCL7A	TSS
-chr13	46957278	46963342	C13orf18	TSS
-chr14	69257848	69259739	ZFP36L1	TSS
-chr14	94940587	94942549	SERPINA9	TSS
-chr14	96179535	96180366	TCL1A	TSS
-chr15	59658991	59671152	MYO1E	TSS
-chr16	10970795	10975465	CIITA	TSS
-chr16	11347512	11350007	SOCS1	intron-1
-chr16	27322895	27329423	IL4R	TSS
-chr16	85931918	85933977	IRF8	TSS
-chr17	3597616	3599572	P2RX5	TSS
-chr17	56407732	56410140	TSPOAP1	intergenic
-chr17	75424734	75471471	SEPT9	intronic
-chr18	60796984	60814103	BCL2	intronic
-chr18	60982728	60988342	BCL2	TSS
-chr19	10340142	10341764	S1PR2	TSS
-chr19	16434978	16439011	KLF2	TSS
-chr19	19279635	19281441	MEF2B	TSS
-chr2	60773789	60783486	BCL11A	TSS
-chr2	96808901	96811913	DUSP2	intron-1
-chr2	111874854	111882174	BCL2L11	TSS
-chr2	136874728	136875461	CXCR4	intronic
-chr20	46128611	46138099	NCOA3	TSS
-chr20	49120482	49140477	PTPN1	TSS
-chr21	26934372	26937651	MIR155HG	TSS
-chr22	23229554	23232042	IGLL5	TSS
-chr3	16546433	16556786	RFTN1	TSS
-chr3	32020518	32024930	OSBPL10	TSS-1
-chr3	71104986	71119518	FOXP1	TSS-4
-chr3	71176427	71180960	FOXP1	TSS-3
-chr3	71350633	71357665	FOXP1	TSS-2
-chr3	71503103	71504870	FOXP1	intron-1b
-chr3	71550753	71551478	FOXP1	intron-1a
-chr3	71626341	71635648	FOXP1	TSS-1
-chr3	101546669	101547704	NFKBIZ	TSS-1
-chr3	101568239	101569274	NFKBIZ	TSS-2
-chr3	183269360	183274139	KLHL6	TSS
-chr3	186737482	186741455	ST6GAL1	TSS-2
-chr3	186739628	186740875	ST6GAL1	TSS-1
-chr3	186781067	186784816	ST6GAL1	intronic-2
-chr3	186783199	186784291	ST6GAL1	intronic-1
-chr3	187458526	187464632	BCL6	TSS
-chr3	187771678	187982852	LPP	TSS-1
-chr3	188377178	188491248	LPP	intronic
-chr4	40193105	40204231	RHOH	TSS
-chr5	88131209	88206620	MEF2C	TSS
-chr5	131823933	131826458	IRF1	TSS
-chr5	149790977	149792349	CD74	TSS
-chr5	158500476	158532769	EBF1	TSS
-chr6	390572	394093	IRF4	TSS
-chr6	14118026	14120025	CD83	TSS
-chr6	31548325	31550717	LTB	intron-1
-chr6	37138104	37139804	PIM1	TSS
-chr6	90981034	91016134	BACH2	TSS
-chr6	134487960	134499859	SGK1	TSS-1
-chr7	5568297	5570856	ACTB	TSS
-chr8	11347723	11355318	BLK	TSS
-chr8	128748352	128749427	MYC	TSS
-chr9	37021323	37043181	PAX5	TSS-2
-chr9	37023396	37027663	PAX5	intron-1
-chr9	37032576	37037704	PAX5	TSS-1
-chr9	37192080	37207549	ZCCHC7	intron-3
-chr9	37276495	37342191	ZCCHC7	intronic
-chr9	37282092	37309161	ZCCHC7	intron-2
-chr9	37323113	37340687	ZCCHC7	intron-1
-chr9	37369209	37372160	PAX5	distal-enhancer-1
-chr9	37382267	37385854	PAX5	distal-enhancer-2
-chr9	37395932	37409239	PAX5	distal-enhancer-3
-chr9	37423010	37425279	GRHPR	TSS
-chrX	12993308	12994511	TMSB4X	intronic
-chrX	48774756	48776255	PIM2	TSS
+chr_name	hg19_start	hg19_end	gene	region	regulatory_comment
+chr1	6661482	6662702	KLHL21	TSS	NA
+chr1	23885584	23885835	ID3	TSS	NA
+chr1	28832551	28836339	RCC1	TSS	NA
+chr1	31229012	31232011	LAPTM5	TSS	NA
+chr1	150550814	150552135	MCL1	intron	NA
+chr2	88904839	88909096	EIF2Ak3	intron	NA
+chr2	88925456	88927581	EIF2Ak3	TSS	NA
+chr2	232572640	232574297	PTMA	TSS	NA
+chr2	157669490	157671299	FCRL3	TSS	NA
+chr1	203274698	203275778	BTG2	intron	active_promoter
+chr1	206285239	206288105	RHEX	TSS	NA
+chr1	226864857	226873452	ITPKB	intron	weak_enhancer
+chr1	226920563	226927885	ITPKB	TSS	active_promoter
+chr1	226921088	226927982	ITPKB	intron-1	enhancer
+chr2	60773789	60783486	BCL11A	TSS	active_promoter
+chr2	96808901	96811913	DUSP2	intron-1	enhancer
+chr2	111874854	111882174	BCL2L11	TSS	active_promoter
+chr2	136874728	136875461	CXCR4	intron	weak_promoter
+chr3	16546433	16556786	RFTN1	TSS	active_promoter-strong_enhancer
+chr3	32020518	32024930	OSBPL10	TSS-1	active_promoter
+chr3	71104986	71119518	FOXP1	TSS-4	active_promoter
+chr3	71176427	71180960	FOXP1	TSS-3	active_promoter
+chr3	71350633	71357665	FOXP1	TSS-2	active_promoter
+chr3	71503103	71504870	FOXP1	intron-1b	intron
+chr3	71550753	71551478	FOXP1	intron-1a	intron
+chr3	71626341	71635648	FOXP1	TSS-1	active_promoter
+chr3	101546669	101547704	NFKBIZ	TSS-1	active_promoter
+chr3	101568239	101569274	NFKBIZ	TSS-2	active_promoter
+chr3	183269360	183274139	KLHL6	TSS	active_promoter-strong_enhancer
+chr3	186647943	186650177	ST6GAL1	TSS-2	NA
+chr3	186739628	186740875	ST6GAL1	TSS-1	active_promoter
+chr3	186709494	186721749	ST6GAL1	intron-2	NA
+chr3	186782522	186784438	ST6GAL1	intron-1	strong_enhancer
+chr3	187458526	187464632	BCL6	TSS	NA
+chr3	187615533	187625659	BCL6	Intergenic-1	NA
+chr3	187625659	187638101	BCL6	Intergenic-2	NA
+chr3	187657518	187665996	BCL6	Intergenic-3	NA
+chr3	187675741	187690717	BCL6	Intergenic-4	NA
+chr3	187690717	187705000	BCL6	Intergenic-5	NA
+chr3	187771678	187801626	LPP	TSS-1	NA
+chr3	187801627	187813568	LPP	TSS-2	NA
+chr3	187813569	187955254	LPP	TSS-3	NA
+chr3	187955255	187969220	LPP	TSS-4	NA
+chr3	188261523	188327097	LPP	TSS-5	NA
+chr3	188377178	188401951	LPP	intron-1	NA
+chr3	188401952	188461879	LPP	intron-2	NA
+chr3	188461880	188474683	LPP	intron-3	NA
+chr3	188474684	188491248	LPP	intron-4	NA
+chr4	25861094	25868788	SEL1L3	TSS	NA
+chr4	40193105	40204231	RHOH	TSS	active_promoter
+chr5	88131209	88174373	MEF2C	TSS-3	active_promoter
+chr5	88174374	88182243	MEF2C	TSS-2	active_promoter
+chr5	88182244	88206620	MEF2C	TSS-1	active_promoter
+chr5	131823933	131826458	IRF1	TSS	active_promoter
+chr5	149790977	149792349	CD74	TSS	active_promoter
+chr5	158500476	158532769	EBF1	TSS	active_promoter
+chr6	390572	394093	IRF4	TSS	active_promoter
+chr6	14118026	14120025	CD83	TSS	active_promoter-strong_enhancer
+chr6	25861094	25868788	RNF144B	TSS	NA
+chr6	31548325	31550717	LTB	intron-1	enhancer
+chr6	37138104	37139804	PIM1	TSS	active_promoter
+chr6	90981034	91016134	BACH2	TSS	NA
+chr6	90059730	90063703	UBE2J1	TSS	NA
+chr6	134487960	134499859	SGK1	TSS-1	active_promoter
+chr6	159237903	159240216	EZR	TSS	NA
+chr7	5568297	5570856	ACTB	TSS	active_promoter
+chr8	11347723	11355318	BLK	TSS	strong_enhancer
+chr8	128748352	128749427	MYC	TSS	active_promoter
+chr9	37023396	37027663	PAX5	intron-1	intron
+chr9	37029849	37037154	PAX5	TSS-1	active_promoter
+chr9	37192080	37207549	ZCCHC7	intron-4	intron
+chr9	37275952	37306152	ZCCHC7	intron-2	intron
+chr9	37312655	37328260	ZCCHC7	intron-3	intron
+chr9	37329706	37340398	ZCCHC7	intron-1	intron
+chr9	37369209	37372160	PAX5	distal-enhancer-1	enhancer
+chr9	37382267	37385854	PAX5	distal-enhancer-2	enhancer
+chr9	37395932	37409239	PAX5	distal-enhancer-3	enhancer
+chr9	37423010	37425279	GRHPR	TSS	active_promoter
+chr9	130740362	130744800	FAM102A	TSS	NA
+chr9	132800698	132805491	FNBP1	TSS	NA
+chr10	127578912	127591133	FANK1	TSS	active_promoter
+chr11	9595246	9599502	WEE1	intron	NA
+chr11	35156769	35164248	CD44	TSS	active_promoter
+chr11	60223385	60225310	MS4A1	TSS	active_promoter
+chr11	65190369	65192380	NEAT1	ncRNA	enhancer
+chr11	65265237	65268359	MALAT1	ncRNA	enhancer
+chr11	102188170	102190077	BIRC3	TSS	active_promoter
+chr11	111248078	111250224	POU2AF1	TSS	active_promoter
+chr11	118754458	118756514	CXCR5	TSS	NA
+chr11	128339774	128345731	ETS1	intron	enhancer
+chr11	128388492	128394163	ETS1	TSS-2	active_promoter
+chr12	8762328	8765099	ACIDA	TSS	NA
+chr12	11796001	11812968	ETV6	TSS	strong_enhancer
+chr12	25204045	25213569	LRMP	TSS	active_promoter
+chr12	92537999	92539598	BTG1	TSS	active_promoter
+chr12	113492311	113497546	DTX1	TSS	NA
+chr12	122456912	122464036	BCL7A	TSS	poised_promoter
+chr13	46957278	46963342	C13orf18	TSS	active_promoter
+chr13	75982519	75986015	TBC1D4	intron	NA
+chr14	69257848	69259739	ZFP36L1	TSS	active_promoter
+chr14	94940587	94942549	SERPINA9	TSS	NA
+chr14	96179535	96180366	TCL1A	TSS	active_promoter
+chr15	59658991	59671152	MYO1E	TSS	NA
+chr16	10970795	10975465	CIITA	TSS	active_promoter-strong_enhancer
+chr16	11347512	11350007	SOCS1	intron-1	enhancer
+chr16	27322895	27329423	IL4R	TSS	active_promoter
+chr16	85931918	85933977	IRF8	TSS	active_promoter
+chr17	3597616	3599572	P2RX5	TSS	active_promoter
+chr17	37928959	37940119	IKZF3	TSS-1	NA
+chr17	38015776	38024832	IKZF3	TSS-2	NA
+chr17	56407732	56410140	TSPOAP1	intergenic	enhancer
+chr17	75424734	75440956	SEPTIN9	intron-1	active_promoter
+chr17	75443766	75451177	SEPTIN9	intron-2	active_promoter
+chr17	75453203	75471471	SEPTIN9	intron-3	active_promoter
+chr17	79478289	79479959	ACTG1	intron	NA
+chr18	60796984	60814103	BCL2	intron	strong_enhancer
+chr18	60982728	60988342	BCL2	TSS	active_promoter
+chr19	10340142	10341764	S1PR2	TSS	active_promoter
+chr19	16434978	16439011	KLF2	TSS	intron
+chr19	19279635	19281441	MEF2B	TSS	active_promoter
+chr20	46128611	46138099	NCOA3	TSS	active_promoter
+chr20	49120482	49140477	PTPN1	TSS	active_promoter
+chr21	26934372	26937651	MIR155HG	TSS	active_promoter
+chr22	23229554	23232042	IGLL5	TSS	NA
+chr22	23522060	23528313	BCR	TSS	NA
+chr22	29194943	29197461	XBP1	TSS	NA
+chrX	12993308	12994511	TMSB4X	intron	active_promoter
+chrX	42800580	42804184	DDX3X	intergenic	NA
+chrX	48774756	48776255	PIM2	TSS	active_promoter