miga-base 0.7.26.0 → 1.0.0.1

data/utils/enveomics/Tests/hiv_mix-hiv1.blast.tsv.lim

@@ -0,0 +1 @@
+gi|4558520|gb|AF033819.3|	1	9182

data/utils/enveomics/Tests/hiv_mix-hiv1.blast.tsv.rec

@@ -0,0 +1,233 @@
+8528	8627	97.03	169	2
+4500	4599	94.00	152	3
+554	653	96.00	161	6
+5564	5663	93.07	147	9
+1542	1641	95.00	158	12
+4743	4842	94.00	152	14
+2565	2664	94.12	154	17
+7723	7822	96.00	161	18
+2461	2560	94.12	154	19
+7817	7913	96.94	163	23
+8553	8649	96.91	163	25
+214	310	96.91	163	28
+2602	2701	94.12	154	33
+2086	2185	97.00	169	34
+3397	3496	94.00	150	35
+1333	1430	93.94	148	40
+6501	6600	97.03	169	44
+673	772	96.00	163	46
+6090	6189	95.05	158	47
+316	415	96.00	163	48
+4830	4929	97.00	169	49
+3925	4024	95.00	156	50
+6472	6571	95.05	158	51
+5396	5495	95.00	158	52
+958	1057	96.00	163	55
+6744	6841	93.94	148	56
+5974	6073	96.04	163	58
+5867	5966	94.06	152	60
+5438	5536	96.97	167	63
+7137	7236	97.03	169	65
+5002	5101	96.04	163	66
+8509	8606	94.90	154	68
+7966	8064	97.98	171	70
+1900	1999	93.14	148	71
+7554	7653	93.07	147	72
+7408	7507	94.06	152	73
+9071	9170	96.00	163	75
+2	85	95.24	134	75
+3295	3394	96.00	161	76
+68	167	94.12	152	79
+2903	3002	94.00	150	81
+7980	8079	94.00	150	85
+3125	3222	96.00	161	86
+8312	8411	94.06	152	87
+3810	3905	97.92	167	91
+8816	8915	95.00	156	97
+3465	3564	95.00	158	100
+6573	6672	96.04	163	109
+7433	7533	93.20	148	110
+8530	8629	95.00	156	112
+1002	1097	94.85	150	113
+109	208	95.00	156	114
+4311	4409	95.96	161	116
+2592	2688	96.91	161	117
+4431	4530	94.06	152	122
+6772	6868	96.91	163	123
+869	968	94.00	150	125
+8525	8624	98.00	174	126
+4762	4859	93.88	148	127
+3092	3191	97.00	169	128
+2768	2865	97.96	171	132
+3781	3880	94.00	150	138
+390	486	95.88	158	143
+7602	7699	94.90	154	147
+4126	4225	93.00	145	148
+6974	7073	97.00	169	151
+872	968	97.94	169	152
+6321	6420	95.00	158	153
+7890	7989	96.00	163	157
+7250	7348	97.98	172	159
+6863	6962	93.14	148	164
+4252	4351	96.00	161	165
+5421	5520	94.00	152	166
+1997	2096	94.00	150	167
+6221	6320	95.00	156	168
+4212	4311	99.00	178	169
+161	259	95.05	158	170
+1836	1935	96.00	161	172
+4646	4745	93.00	147	173
+2246	2345	95.00	156	175
+4141	4240	92.08	141	177
+7714	7813	94.12	152	181
+2761	2860	95.00	156	185
+2329	2428	96.00	161	189
+951	1050	94.00	152	191
+6926	7025	95.00	158	193
+3092	3190	92.00	139	195
+7058	7157	94.00	150	196
+8895	8994	95.00	156	198
+5428	5527	95.05	156	199
+6003	6102	94.00	150	200
+704	800	96.91	163	205
+6180	6279	96.08	165	206
+1441	1540	96.00	163	207
+4364	4462	94.00	150	210
+3275	3374	95.05	158	211
+2430	2529	97.00	169	213
+2594	2693	95.05	158	215
+3876	3975	97.00	167	217
+1203	1302	96.00	161	219
+2220	2319	98.00	174	221
+9053	9152	95.00	156	223
+2	67	93.94	99.0	223
+5239	5338	95.00	158	225
+2780	2879	94.06	152	228
+423	521	92.93	145	230
+273	372	95.00	158	231
+4097	4196	95.00	158	233
+7224	7323	93.00	147	234
+8474	8573	97.00	169	237
+5615	5714	96.00	163	238
+8804	8900	95.88	158	239
+6719	6818	95.05	158	240
+5362	5458	94.85	150	242
+866	965	94.06	152	251
+8442	8541	92.00	139	253
+7073	7172	94.06	152	254
+1580	1679	91.00	135	255
+4603	4700	95.92	158	256
+519	618	97.00	167	260
+3213	3312	97.00	169	261
+8968	9067	96.00	163	264
+6509	6605	97.94	169	270
+7649	7748	96.00	161	271
+8332	8431	93.27	150	273
+7708	7807	94.06	152	274
+6755	6854	97.00	167	275
+8704	8803	97.00	169	277
+4942	5041	96.00	161	278
+926	1025	96.00	161	280
+213	312	94.06	152	284
+5883	5980	96.94	165	285
+3427	3526	93.07	147	287
+6898	6997	97.03	169	288
+493	592	95.00	158	290
+5247	5344	94.95	154	295
+1142	1241	96.04	163	297
+837	935	97.00	167	298
+5087	5186	93.00	145	301
+9019	9115	94.90	152	302
+8527	8621	96.88	159	303
+9067	9166	95.00	156	305
+2	81	95.00	124	305
+3165	3264	94.00	152	307
+2303	2402	95.05	158	313
+6553	6652	94.00	150	315
+4703	4802	96.00	163	316
+5364	5463	96.00	163	319
+4306	4404	97.98	172	320
+2971	3070	95.05	158	321
+3652	3751	97.03	169	322
+8972	9071	93.00	145	323
+7302	7401	96.00	161	325
+1543	1642	94.06	150	326
+6034	6133	98.00	174	332
+4963	5054	93.48	135	333
+5103	5202	94.06	152	335
+1002	1097	92.71	139	336
+3007	3106	93.14	148	338
+5658	5755	95.92	158	339
+830	929	94.06	152	342
+6979	7078	93.07	147	343
+3926	4025	94.06	152	344
+2047	2145	95.96	159	345
+7952	8051	95.00	158	353
+9018	9111	96.81	158	354
+5508	5607	91.09	135	355
+5710	5809	93.07	147	362
+6296	6395	94.00	152	363
+6554	6653	97.00	169	368
+7297	7396	98.00	174	369
+31	129	96.00	161	370
+9116	9182	95.59	108	370
+6688	6787	93.00	147	371
+5769	5868	96.04	163	372
+5461	5560	96.00	163	376
+1438	1537	96.00	161	377
+3870	3969	97.00	167	380
+6937	7036	95.00	158	385
+1765	1861	96.94	163	387
+7150	7249	95.00	158	388
+6866	6962	97.94	169	389
+678	777	94.06	152	393
+2441	2540	95.00	156	396
+5634	5733	95.00	156	404
+5456	5555	96.00	161	405
+6463	6562	95.00	158	407
+122	221	95.05	158	408
+4519	4618	94.00	152	412
+3679	3778	95.00	158	413
+8626	8725	94.00	152	416
+6577	6676	93.00	147	417
+8543	8639	95.88	156	418
+8753	8852	96.00	161	420
+2224	2323	93.07	147	421
+6645	6744	93.00	145	422
+3892	3991	92.00	139	425
+3599	3698	97.00	169	426
+1408	1508	96.04	165	428
+3341	3440	97.00	167	429
+8316	8415	95.00	156	436
+1809	1908	96.00	163	438
+7770	7869	97.00	169	439
+7250	7349	97.00	169	440
+5590	5688	93.94	150	441
+8475	8574	98.02	174	443
+2139	2237	95.96	161	446
+4074	4173	95.00	158	448
+5269	5367	95.00	156	450
+7237	7336	96.00	161	451
+2461	2559	94.06	152	454
+1168	1267	95.05	158	455
+3083	3178	95.88	156	457
+5129	5228	99.00	180	463
+8590	8684	96.84	159	464
+2677	2776	93.20	148	465
+8332	8431	95.00	156	466
+7692	7791	96.04	163	470
+51	150	95.00	158	471
+1397	1496	98.00	172	472
+42	141	94.00	152	475
+2117	2216	96.00	161	476
+5481	5579	93.94	147	482
+7809	7908	95.00	158	485
+4991	5089	94.00	148	486
+9059	9158	97.00	167	487
+2	73	95.83	115	487
+133	232	95.10	158	488
+6966	7065	95.10	159	490
+6296	6390	96.84	158	492
+4728	4827	97.00	169	495
+4205	4304	96.08	165	498
+6833	6932	98.00	172	499

data/utils/enveomics/Tests/phyla_counts.tsv

@@ -0,0 +1,10 @@
+OTU	Site A	Site B	Site C	Site D
+Actinobacteria	22	1	2	22
+Dehalococcoidia	44	3	2	44
+Fusobacteria	1	4	2	32
+Nitrospirae	32	3	6	1
+Planctomycetes	11	2	1	11
+Proteobacteria	77	25	38	7
+Spirochaetes	4	11	7	3
+Thermotogae	2	11	1	2
+unclassified	1	1	1	1

data/utils/enveomics/Tests/primate_lentivirus.ogs

@@ -0,0 +1,11 @@
+hiv1	hiv2	siv
+lcl|AF033819.3_prot_AAC82593.1_1	lcl|NC_001722.1_prot_NP_056837.1_2	lcl|NC_001549.1_prot_NP_054369.1_2
+lcl|AF033819.3_prot_AAC82598.2_2	lcl|NC_001722.1_prot_NP_663784.1_1	lcl|NC_001549.1_prot_NP_687035.1_1
+lcl|AF033819.3_prot_AAC82594.1_3	lcl|NC_001722.1_prot_NP_056839.1_3	lcl|NC_001549.1_prot_NP_054370.1_3
+lcl|AF033819.3_prot_AAC82595.1_4	lcl|NC_001722.1_prot_NP_056841.1_5	lcl|NC_001549.1_prot_NP_054371.1_4
+lcl|AF033819.3_prot_AAC82591.1_5	lcl|NC_001722.1_prot_NP_056842.1_6	-
+lcl|AF033819.3_prot_AAC82592.1_6	lcl|NC_001722.1_prot_NP_056843.1_7	-
+lcl|AF033819.3_prot_AAD20388.1_7	-	-
+lcl|AF033819.3_prot_AAC82596.1_8	lcl|NC_001722.1_prot_NP_056844.1_8	lcl|NC_001549.1_prot_NP_054372.1_5
+lcl|AF033819.3_prot_AAC82597.1_9	lcl|NC_001722.1_prot_NP_056845.1_9	lcl|NC_001549.1_prot_NP_054373.1_6
+-	lcl|NC_001722.1_prot_NP_056840.1_4	-

data/utils/enveomics/Tests/primate_lentivirus.rbm/hiv1-hiv1.rbm

@@ -0,0 +1,9 @@
+lcl|AF033819.3_prot_AAC82593.1_1	lcl|AF033819.3_prot_AAC82593.1_1	100.00	500	0	0	1	500	1	500	0.0	1044	500	500
+lcl|AF033819.3_prot_AAC82598.2_2	lcl|AF033819.3_prot_AAC82598.2_2	100.00	1003	0	0	1	1003	1	1003	0.0	2062	1003	1003
+lcl|AF033819.3_prot_AAC82594.1_3	lcl|AF033819.3_prot_AAC82594.1_3	100.00	192	0	0	1	192	1	192	6e-147	400	192	192
+lcl|AF033819.3_prot_AAC82595.1_4	lcl|AF033819.3_prot_AAC82595.1_4	100.00	78	0	0	1	78	1	78	4e-56	161	78	78
+lcl|AF033819.3_prot_AAC82591.1_5	lcl|AF033819.3_prot_AAC82591.1_5	100.00	86	0	0	1	86	1	86	6e-63	179	86	86
+lcl|AF033819.3_prot_AAC82592.1_6	lcl|AF033819.3_prot_AAC82592.1_6	100.00	116	0	0	1	116	1	116	1e-82	231	116	116
+lcl|AF033819.3_prot_AAD20388.1_7	lcl|AF033819.3_prot_AAD20388.1_7	100.00	82	0	0	1	82	1	82	2e-54	157	82	82
+lcl|AF033819.3_prot_AAC82596.1_8	lcl|AF033819.3_prot_AAC82596.1_8	100.00	856	0	0	1	856	1	856	0.0	1770	856	856
+lcl|AF033819.3_prot_AAC82597.1_9	lcl|AF033819.3_prot_AAC82597.1_9	100.00	123	0	0	1	123	1	123	1e-92	257	123	123

data/utils/enveomics/Tests/primate_lentivirus.rbm/hiv1-hiv2.rbm

@@ -0,0 +1,8 @@
+lcl|AF033819.3_prot_AAC82593.1_1	lcl|NC_001722.1_prot_NP_056837.1_2	58.19	452	178	7	1	448	1	445	0.0	526	500	521
+lcl|AF033819.3_prot_AAC82598.2_2	lcl|NC_001722.1_prot_NP_663784.1_1	58.24	953	397	1	42	994	498	1449	0.0	1169	1003	1550
+lcl|AF033819.3_prot_AAC82594.1_3	lcl|NC_001722.1_prot_NP_056839.1_3	33.75	160	91	6	2	153	4	156	2e-20	75.9	192	215
+lcl|AF033819.3_prot_AAC82595.1_4	lcl|NC_001722.1_prot_NP_056841.1_5	40.79	76	45	0	2	77	7	82	1e-18	65.9	78	87
+lcl|AF033819.3_prot_AAC82591.1_5	lcl|NC_001722.1_prot_NP_056842.1_6	67.74	31	9	1	18	47	46	76	9e-12	47.8	86	130
+lcl|AF033819.3_prot_AAC82592.1_6	lcl|NC_001722.1_prot_NP_056843.1_7	46.43	28	14	1	8	34	5	32	0.004	25.0	116	103
+lcl|AF033819.3_prot_AAC82596.1_8	lcl|NC_001722.1_prot_NP_056844.1_8	34.47	879	492	22	25	856	19	860	3e-146	447	856	860
+lcl|AF033819.3_prot_AAC82597.1_9	lcl|NC_001722.1_prot_NP_056845.1_9	56.90	58	25	0	66	123	98	155	2e-22	80.1	123	257

data/utils/enveomics/Tests/primate_lentivirus.rbm/hiv1-siv.rbm

@@ -0,0 +1,6 @@
+lcl|AF033819.3_prot_AAC82593.1_1	lcl|NC_001549.1_prot_NP_054369.1_2	52.67	524	208	9	1	496	1	512	0.0	518	500	513
+lcl|AF033819.3_prot_AAC82598.2_2	lcl|NC_001549.1_prot_NP_687035.1_1	60.25	966	373	6	36	994	487	1448	0.0	1196	1003	1472
+lcl|AF033819.3_prot_AAC82594.1_3	lcl|NC_001549.1_prot_NP_054370.1_3	38.71	155	87	7	2	152	4	154	1e-25	90.9	192	219
+lcl|AF033819.3_prot_AAC82595.1_4	lcl|NC_001549.1_prot_NP_054371.1_4	30.43	69	40	3	11	73	21	87	4e-07	34.3	78	118
+lcl|AF033819.3_prot_AAC82596.1_8	lcl|NC_001549.1_prot_NP_054372.1_5	36.16	874	464	26	34	855	22	853	5e-154	467	856	854
+lcl|AF033819.3_prot_AAC82597.1_9	lcl|NC_001549.1_prot_NP_054373.1_6	60.34	58	23	0	66	123	82	139	5e-21	75.9	123	223

data/utils/enveomics/Tests/primate_lentivirus.rbm/hiv2-hiv2.rbm

@@ -0,0 +1,9 @@
+lcl|NC_001722.1_prot_NP_663784.1_1	lcl|NC_001722.1_prot_NP_663784.1_1	100.00	1550	0	0	1	1550	1	1550	0.0	3210	1550	1550
+lcl|NC_001722.1_prot_NP_056837.1_2	lcl|NC_001722.1_prot_NP_056837.1_2	100.00	521	0	0	1	521	1	521	0.0	1077	521	521
+lcl|NC_001722.1_prot_NP_056839.1_3	lcl|NC_001722.1_prot_NP_056839.1_3	100.00	215	0	0	1	215	1	215	5e-164	446	215	215
+lcl|NC_001722.1_prot_NP_056840.1_4	lcl|NC_001722.1_prot_NP_056840.1_4	100.00	113	0	0	1	113	1	113	6e-85	237	113	113
+lcl|NC_001722.1_prot_NP_056841.1_5	lcl|NC_001722.1_prot_NP_056841.1_5	100.00	87	0	0	1	87	1	87	5e-63	179	87	87
+lcl|NC_001722.1_prot_NP_056842.1_6	lcl|NC_001722.1_prot_NP_056842.1_6	100.00	130	0	0	1	130	1	130	4e-94	262	130	130
+lcl|NC_001722.1_prot_NP_056843.1_7	lcl|NC_001722.1_prot_NP_056843.1_7	100.00	103	0	0	1	103	1	103	2e-72	204	103	103
+lcl|NC_001722.1_prot_NP_056844.1_8	lcl|NC_001722.1_prot_NP_056844.1_8	100.00	860	0	0	1	860	1	860	0.0	1799	860	860
+lcl|NC_001722.1_prot_NP_056845.1_9	lcl|NC_001722.1_prot_NP_056845.1_9	100.00	257	0	0	1	257	1	257	0.0	536	257	257

data/utils/enveomics/Tests/primate_lentivirus.rbm/hiv2-siv.rbm

@@ -0,0 +1,6 @@
+lcl|NC_001722.1_prot_NP_663784.1_1	lcl|NC_001549.1_prot_NP_687035.1_1	56.99	1474	601	10	1	1458	1	1457	0.0	1697	1550	1472
+lcl|NC_001722.1_prot_NP_056837.1_2	lcl|NC_001549.1_prot_NP_054369.1_2	57.39	528	203	7	1	521	1	513	0.0	593	521	513
+lcl|NC_001722.1_prot_NP_056839.1_3	lcl|NC_001549.1_prot_NP_054370.1_3	44.97	169	88	5	1	167	1	166	6e-41	132	215	219
+lcl|NC_001722.1_prot_NP_056841.1_5	lcl|NC_001549.1_prot_NP_054371.1_4	35.21	71	42	2	17	83	22	92	3e-11	46.2	87	118
+lcl|NC_001722.1_prot_NP_056844.1_8	lcl|NC_001549.1_prot_NP_054372.1_5	44.90	873	401	14	25	859	23	853	0.0	714	860	854
+lcl|NC_001722.1_prot_NP_056845.1_9	lcl|NC_001549.1_prot_NP_054373.1_6	41.74	230	108	5	1	224	1	210	2e-50	158	257	223

data/utils/enveomics/Tests/primate_lentivirus.rbm/siv-siv.rbm

@@ -0,0 +1,6 @@
+lcl|NC_001549.1_prot_NP_687035.1_1	lcl|NC_001549.1_prot_NP_687035.1_1	100.00	1472	0	0	1	1472	1	1472	0.0	3045	1472	1472
+lcl|NC_001549.1_prot_NP_054369.1_2	lcl|NC_001549.1_prot_NP_054369.1_2	100.00	513	0	0	1	513	1	513	0.0	1067	513	513
+lcl|NC_001549.1_prot_NP_054370.1_3	lcl|NC_001549.1_prot_NP_054370.1_3	100.00	219	0	0	1	219	1	219	7e-169	459	219	219
+lcl|NC_001549.1_prot_NP_054371.1_4	lcl|NC_001549.1_prot_NP_054371.1_4	100.00	118	0	0	1	118	1	118	3e-87	243	118	118
+lcl|NC_001549.1_prot_NP_054372.1_5	lcl|NC_001549.1_prot_NP_054372.1_5	100.00	854	0	0	1	854	1	854	0.0	1776	854	854
+lcl|NC_001549.1_prot_NP_054373.1_6	lcl|NC_001549.1_prot_NP_054373.1_6	100.00	223	0	0	1	223	1	223	5e-170	462	223	223

data/utils/enveomics/build_enveomics_r.bash

@@ -0,0 +1,45 @@
+#!/bin/bash
+
+cd $(dirname -- $0)/enveomics.R
+rm man/*
+echo '
+\name{phyla.counts}
+\docType{data}
+\alias{phyla.counts}
+\title{Counts of microbial phyla in four sites}
+\description{
+  This data set gives the counts of phyla in three different
+  sites.
+}
+\usage{phyla.counts}
+\format{A data frame with 9 rows (phyla) and 4 rows (sites).}
+\keyword{datasets}
+' > man/phyla.counts.Rd
+echo '
+\name{growth.curves}
+\docType{data}
+\alias{growth.curves}
+\title{Bacterial growth curves for three Escherichia coli mutants}
+\description{
+  This data set provides time (first column) and three triplicated growth
+  curves as optical density at 600nm (OD_600nm) for different mutants of E.
+  coli.
+}
+\usage{growth.curves}
+\format{A data frame with 16 rows (times) and 10 rows (times and OD_600nm).}
+\keyword{datasets}
+' > man/growth.curves.Rd
+echo 'roxygen2::roxygenise();' | R --vanilla
+#echo "
+#library(inlinedocs)
+#package.skeleton.dx('./');
+#" | R --vanilla
+#cat man/enveomics.R-package.Rd | tr -d '\r' \
+#  | grep -v '^}$' | grep -v '^\\author{' \
+#  | grep -v '^Maintainer' \
+#  | perl -pe 's/^\\keyword/}\n\\author{Luis M. Rodriguez-R <lmrodriguezr\@gmail.com> [aut, cre]}\n\n\\keyword/' \
+#  | perl -lwe '$/=\0; $_=<>; s/^\\details{\n+([^}].*\n+)*}\n+//mg; print' \
+#  > o && mv o man/enveomics.R-package.Rd
+#[[ ! -d inst/doc ]] && mkdir -p inst/doc
+#pandoc -o inst/doc/enveomics.R.pdf -f markdown_github README.md
+

data/utils/enveomics/enveomics.R/DESCRIPTION

@@ -0,0 +1,31 @@
+Package: enveomics.R
+Version: 1.8.0
+Authors@R: c(person("Luis M.","Rodriguez-R",role=c("aut","cre"),
+	   email="lmrodriguezr@gmail.com"))
+Title: Various Utilities for Microbial Genomics and Metagenomics
+Description: A collection of functions for microbial ecology and other
+   applications of genomics and metagenomics. Companion package for the
+   Enveomics Collection (Rodriguez-R, L.M. and Konstantinidis, K.T., 2016
+   <DOI:10.7287/peerj.preprints.1900v1>).
+Author: Luis M. Rodriguez-R [aut, cre]
+Maintainer: Luis M. Rodriguez-R <lmrodriguezr@gmail.com>
+URL: http://enve-omics.ce.gatech.edu/enveomics
+Depends:
+   R (>= 2.9),
+   stats,
+   methods,
+   parallel,
+   fitdistrplus,
+   sn,
+   investr
+Suggests:
+   tools,
+   vegan,
+   ape,
+   picante,
+   gplots,
+   optparse
+License: Artistic-2.0
+LazyData: yes
+Encoding: UTF-8
+RoxygenNote: 7.0.2

data/utils/enveomics/enveomics.R/NAMESPACE

@@ -0,0 +1,39 @@
+import(methods)
+import(parallel)
+import(fitdistrplus)
+import(sn)
+importFrom("investr", "predFit")
+importFrom("grDevices", "boxplot.stats", "col2rgb", "grey", "rainbow",
+  "rgb")
+importFrom("graphics", "abline", "axis", "barplot", "hist", "image",
+  "layout", "legend", "lines", "par", "points", "polygon",
+  "rect", "text")
+importFrom("stats", "as.dist", "cophenetic", "cor", "median",
+  "quantile", "runif", "smooth.spline", "nls", "nls.control", "qnorm",
+  "dnorm", "kmeans")
+importFrom("utils", "head", "read.table", "setTxtProgressBar", "tail",
+  "txtProgressBar")
+exportClasses(enve.RecPlot2, enve.RecPlot2.Peak, enve.TRIBS, enve.TRIBStest,
+  enve.GrowthCurve)
+S3method(plot, enve.RecPlot2)
+S3method(plot, enve.TRIBS)
+S3method(plot, enve.TRIBStest)
+S3method(plot, enve.GrowthCurve)
+S3method(summary, enve.TRIBS)
+S3method(summary, enve.TRIBStest)
+S3method(summary, enve.GrowthCurve)
+export(
+  enve.cliopts, enve.col.alpha,
+  enve.barplot, enve.recplot, enve.prune.dist,
+  enve.tribs, plot.enve.TRIBS, summary.enve.TRIBS, enve.tribs.test,
+  plot.enve.TRIBStest, summary.enve.TRIBStest,
+  enve.df2dist, enve.df2dist.group, enve.df2dist.list,
+  enve.recplot2, plot.enve.RecPlot2, enve.recplot2.findPeaks,
+  enve.recplot2.findPeaks.emauto, enve.recplot2.findPeaks.em,
+  enve.recplot2.findPeaks.mower,
+  enve.recplot2.corePeak, enve.recplot2.changeCutoff,
+  enve.recplot2.windowDepthThreshold, enve.recplot2.extractWindows,
+  enve.recplot2.compareIdentities,
+  enve.recplot2.coordinates, enve.recplot2.seqdepth, enve.recplot2.ANIr,
+  enve.growthcurve, plot.enve.GrowthCurve, summary.enve.GrowthCurve,
+  enve.col2alpha, enve.truncate, enve.selvector, enve.prefscore)

data/utils/enveomics/enveomics.R/R/autoprune.R

@@ -0,0 +1,155 @@
+#' Enveomics: Prune Dist
+#' 
+#' Automatically prunes a tree, to keep representatives of each clade.
+#'
+#' @param t A \strong{phylo} object or a path to the Newick file.
+#' @param dist.quantile The quantile of edge lengths.
+#' @param min_dist The minimum distance to allow between two tips.
+#' If not set, \code{dist.quantile} is used instead to calculate it.
+#' @param quiet Boolean indicating if the function must run without output.
+#' @param max_iters Maximum number of iterations.
+#' @param min_nodes_random 
+#' Minimum number of nodes to trigger \emph{tip-pairs} nodes sampling. 
+#' This sampling is less reproducible and more computationally expensive,
+#' but it's the only solution if the cophenetic matrix exceeds \code{2^31-1} 
+#' entries; above that, it cannot be represented in R.
+#' @param random_nodes_frx 
+#' Fraction of the nodes to be sampled if more than \code{min_nodes_random}.
+#'
+#' @return Returns a pruned \strong{phylo} object.
+#'
+#' @author Luis M. Rodriguez-R [aut, cre]
+#'
+#' @export
+
+enve.prune.dist <- function
+(t,
+ dist.quantile=0.25,
+ min_dist,
+ quiet=FALSE,
+ max_iters=100,
+ min_nodes_random=4e4,
+ random_nodes_frx=1
+){
+  if(!requireNamespace("ape", quietly=TRUE))
+    stop('Unavailable ape library.');
+  if(is.character(t)) t <- ape::read.tree(t)
+  if(missing(min_dist)){
+    if(dist.quantile>0){
+      min_dist <- as.numeric(quantile(t$edge.length, dist.quantile));
+    }else{
+      min_dist <- as.numeric(min(t$edge.length[t$edge.length>0]));
+    }
+  }
+  if(!quiet) cat('\nObjective minimum distance: ',min_dist,', initial tips: ',length(t$tip.label),'\n', sep='');
+  round=1;
+  while(round <= max_iters){
+    if(length(t$tip.label) > min_nodes_random){
+      if(!quiet) cat('  | Iter: ',round-1,', Tips: ', length(t$tip.label),
+                     ', reducing tip-pairs.\n', sep='');
+      rnd.nodes <- sample(t$tip.label, length(t$tip.label)*random_nodes_frx);
+      t <- enve.__prune.reduce(t, rnd.nodes, min_dist, quiet);
+    }else{
+      if(!quiet) cat(' Gathering distances...\r');
+      d <- cophenetic(t);
+      diag(d) <- NA;
+      if(!quiet) cat('  | Iter: ',round-1,', Tips: ', length(t$tip.label),
+                     ', Median distance: ', median(d, na.rm=TRUE),
+                     ', Minimum distance: ', min(d, na.rm=TRUE),
+                     '\n', sep='');
+      # Run iteration
+      if(min(d, na.rm=TRUE) < min_dist){
+        t <- enve.__prune.iter(t, d, min_dist, quiet);
+      }else{
+        break;
+      }
+    }
+    round <- round + 1;
+  }
+  return(t);
+}
+
+#' Enveomics: Prune Reduce (Internal Function)
+#' 
+#' Internal function for \code{\link{enve.prune.dist}}.
+#'
+#' @param t A \strong{phylo} object
+#' @param nodes Vector of nodes
+#' @param min_dist Minimum distance
+#' @param quiet If running quietly
+#' 
+#' @author Luis M. Rodriguez-R [aut, cre]
+#'
+#' @export
+
+enve.__prune.reduce <- function
+(t, nodes, min_dist, quiet){
+  if(!quiet) pb <- txtProgressBar(1, length(nodes), style=3);
+  for(i in 1:length(nodes)){
+    node.name <- nodes[i];
+    if(!quiet) setTxtProgressBar(pb, i);
+    # Get node ID
+    node <- which(t$tip.label==node.name);
+    if(length(node)==0) next;
+    # Get parent and distance to parent
+    parent.node <- t$edge[ t$edge[,2]==node, 1];
+    # Get edges to parent
+    parent.edges <- which(t$edge[,1]==parent.node);
+    stopit <- FALSE;
+    for(j in parent.edges){
+      for(k in parent.edges){
+        if(j != k & t$edge[j,2]<length(t$tip.label) & t$edge[k,2]<length(t$tip.label) & sum(t$edge.length[c(j,k)]) < min_dist){
+          t <- ape::drop.tip(t, t$edge[k,2]);
+          stopit <- TRUE;
+          break;
+        }
+      }
+      if(stopit) break;
+    }
+  }
+  if(!quiet) cat('\n');
+  return(t);
+}
+
+#' Enveomics: Prune Iter (Internal Function)
+#' 
+#' Internal function for \code{\link{enve.prune.dist}}.
+#'
+#' @param t A \strong{phylo} object
+#' @param dist Cophenetic distance matrix
+#' @param min_dist Minimum distance
+#' @param quiet If running quietly
+#' 
+#' @author Luis M. Rodriguez-R [aut, cre]
+#'
+#' @export
+
+enve.__prune.iter <- function
+(t,
+ dist,
+ min_dist,
+ quiet){
+  ori_len <- length(t$tip.label);
+  # Prune
+  if(!quiet) pb <- txtProgressBar(1, ncol(dist)-1, style=3);
+  ignore <- c();
+  for(i in 1:(ncol(dist)-1)){
+    if(i %in% ignore) next;
+    for(j in (i+1):nrow(dist)){
+      if(dist[j, i]<min_dist){
+        t <- ape::drop.tip(t, rownames(dist)[j]);
+        ignore <- c(ignore, j);
+        break;
+      }
+    }
+    if(!quiet) setTxtProgressBar(pb, i);
+  }
+  if(!quiet) cat('\n');
+  # Check if it droped tips
+  cur_len <- length(t$tip.label);
+  if(cur_len == ori_len){
+    stop("Internal error: small edge found in tree, with no equivalent in distance matrix.\n");
+  }
+  return(t);
+}
+