1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466
|
# Copyright (C) Tal Galili
#
# This file is part of dendextend.
#
# dendextend is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# dendextend is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# A copy of the GNU General Public License is available at
# http://www.r-project.org/Licenses/
#
#' @title untangle dendrograms
#' @export
#' @rdname untangle
#' @description
#' One untangle function to rule them all.
#'
#' This function untangles dendrogram lists (dendlist),
#' Using various heuristics.
#'
#' @author Tal Galili
#'
#' @param dend1 a dendrogram or a dendlist object
#' @param dend2 A second dendrogram (to untangle against)
#' @param which an integer vector of length 2, indicating
#' which of the trees in the dendlist object should be plotted
#' @param method a character indicating the type of untangle
#' heuristic to use. The options are:
#' ("labels", "ladderize", "random", "step1side", "step2side", "stepBothSides",
#' "DendSer")
#' @param ... passed to the relevant untangle function
#' @details
#' This function wraps all of the untangle functions,
#' in order to make it easier to find our about (and use) them.
#' @return A \link{dendlist}, with two trees after
#' they have been untangled.
#'
#' If the dendlist was originally larger than 2, it will return the original dendlist
#' but with the relevant trees properly rotate.
#'
#' @seealso
#' \link{tanglegram}, \link{untangle_random_search},
#' \link{untangle_step_rotate_1side}, \link{untangle_step_rotate_2side},
#' \link{untangle_DendSer},
#' \link{entanglement}
#' @examples
#' \dontrun{
#' set.seed(23235)
#' ss <- sample(1:150, 10)
#' dend1 <- iris[ss, -5] %>%
#' dist() %>%
#' hclust("com") %>%
#' as.dendrogram()
#' dend2 <- iris[ss, -5] %>%
#' dist() %>%
#' hclust("sin") %>%
#' as.dendrogram()
#' dend12 <- dendlist(dend1, dend2)
#'
#' dend12 %>% tanglegram()
#'
#' untangle(dend1, dend2, method = "random", R = 5) %>% tanglegram()
#'
#' # it works, and we get something different:
#' set.seed(1234)
#' dend12 %>%
#' untangle(method = "random", R = 5) %>%
#' tanglegram()
#'
#' set.seed(1234)
#' # fixes it completely:
#' dend12 %>%
#' untangle(method = "random", R = 5) %>%
#' untangle(method = "step1") %>%
#' tanglegram()
#' # not good enough
#' dend12 %>%
#' untangle(method = "step1") %>%
#' tanglegram()
#' # not good enough
#' dend12 %>%
#' untangle(method = "step2") %>%
#' tanglegram()
#' # How we might wish to use it:
#' set.seed(12777)
#' dend12 %>%
#' untangle(method = "random", R = 1) %>%
#' untangle(method = "step2") %>%
#' tanglegram()
#' }
untangle <- function(dend1, ...) {
UseMethod("untangle")
}
#' @export
#' @rdname untangle
untangle.default <- function(dend1, ...) {
stop("No default function for tanglegram - must use a dendrogram/hclust/phylo object")
}
#' @export
#' @rdname untangle
untangle_labels <- function(dend1, dend2, ...) {
dend2 <- rotate(dend2, labels(dend1))
dendlist(dend1, dend2)
}
#' @export
#' @rdname untangle
untangle.dendrogram <- function(dend1, dend2,
method = c("labels", "ladderize", "random", "step1side", "step2side", "stepBothSides", "DendSer"), ...) {
method <- match.arg(method)
switch(method,
random = untangle_random_search(dend1, dend2, ...),
step1side = untangle_step_rotate_1side(dend1, dend2, ...),
step2side = untangle_step_rotate_2side(dend1, dend2, ...),
stepBothSides = untangle_step_rotate_both_side(dend1, dend2, ...),
DendSer = untangle_DendSer(dendlist(dend1, dend2), ...),
ladderize = ladderize(dendlist(dend1, dend2), ...),
labels = untangle_labels(dend1, dend2, ...)
)
}
#' @export
#' @rdname untangle
untangle.dendlist <- function(dend1,
method = c("labels", "ladderize", "random", "step1side", "step2side", "DendSer"),
which = c(1L, 2L), ...) {
method <- match.arg(method)
the_names <- names(dend1)[which]
untangle_result <- untangle(dend1[[which[1]]], dend1[[which[2]]], method = method, ...)
if (length(dend1) > 2) {
dend1[[which[1]]] <- untangle_result[[1]]
dend1[[which[2]]] <- untangle_result[[2]]
names(dend1) <- the_names
return(dend1)
} else { # no need for all the copying if the list had only two elements in it.
names(untangle_result) <- the_names
return(untangle_result)
}
}
# center <- function(type = c("mean", "median", "trimmed")) {
# print(match.arg(type))
# }
# center(type="tri")
# get("sort")
#' 'shuffle' is a function that randomilly rotates ("shuffles") a tree.
#' a dendrogram leaves order (by means of rotation)
#' @title Random rotation of trees
#' @export
#' @rdname shuffle
#'
#' @description
#' 'shuffle' randomilly rotates ("shuffles") a tree, changing its presentation
#' while preserving its topolgoy.
#' 'shuffle' is based on \link[dendextend]{rotate} and through its methods can
#' work for any of the major tree objects in R (\link{dendrogram}/\link{hclust}/\link[ape]{phylo}).
#'
#' This function is useful in combination with \link{tanglegram} and \link{entanglement}.
#'
#' @param dend a tree object (\link{dendrogram}/\link{hclust}/\link[ape]{phylo})
#' @param which an integer vector for indicating
#' which of the trees in the dendlist object should be plotted
#' default is missing, in which case all the dends in dendlist
#' will be shuffled
#' @param ... Ignored.
#'
#' @return A randomlly rotated tree object
#' @seealso \code{\link{tanglegram}}, \code{\link{entanglement}},
#' \code{\link[dendextend]{rotate}}
#' @examples
#' dend <- USArrests %>%
#' dist() %>%
#' hclust() %>%
#' as.dendrogram()
#' set.seed(234238)
#' dend2 <- shuffle(dend)
#'
#' tanglegram(dend, dend2, margin_inner = 7)
#' entanglement(dend, dend2) # 0.3983
#'
#' # although these ARE the SAME tree:
#' tanglegram(sort(dend), sort(dend2), margin_inner = 7)
shuffle <- function(dend, ...) {
UseMethod("shuffle")
}
#' @export
#' @rdname shuffle
shuffle.default <- function(dend, ...) {
# takes a dendrogram object and shuffles its branches in a random fashion
# n_leaves <- length(labels(dend)) # leaves.value is faster then labels!
n_leaves <- nleaves(dend)
random_weights <- sample(seq_len(n_leaves)) # a random ordaring of 1:n_leaves weights
rotate(dend, random_weights) # since we have a method here for dend/hclust/phylo - this makes this function rather generic...
}
#' @export
#' @rdname shuffle
shuffle.dendrogram <- shuffle.default
#' @export
#' @rdname shuffle
shuffle.dendlist <- function(dend, which, ...) {
# if(T) 1 else 2
# if(F) 1 else 2
# if(F) 1 else
# 2
what_to_shuffle <- if (missing(which)) seq_len(length(dend)) else which
for (i in what_to_shuffle) {
dend[[i]] <- shuffle(dend[[i]])
}
dend
}
#' @export
#' @rdname shuffle
shuffle.hclust <- shuffle.default
#' @export
#' @rdname shuffle
shuffle.phylo <- shuffle.default
#' @title Untangle - random search
#' @export
#' @description
#' Searches for two untangled dendrogram by randomlly shuflling them and each
#' time checking if their entanglement was improved.
#'
#' @param dend1 a tree object (of class dendrogram/hclust/phylo).
#' @param dend2 a tree object (of class dendrogram/hclust/phylo).
#' @param R numeric (default is 100). The number of shuffles to perform.
#' @param L the distance norm to use for measuring the distance between the
#' two trees. It can be any positive number, often one will want to
#' use 0, 1, 1.5, 2 (see 'details' for more).
#' It is passed to \link{entanglement}.
#' @param leaves_matching_method a character scalar passed to \link{entanglement}.
#' It can be either "order" or "labels" (default). If using "labels",
#' then we use the labels for matching the leaves order value.
#' And if "order" then we use the old leaves order value for matching the
#' leaves order value.
#'
#' Using "order" is faster, but "labels" is safer. "order" will assume that
#' the original two trees had their labels and order values MATCHED.
#'
#' Hence, it is best to make sure that the trees used here have the same labels
#' and the SAME values matched to these values - and then use "order" (for
#' fastest results).
#'
#' If "order" is used, the function first calls \link{match_order_by_labels}
#' in order to make sure that the two trees have their labels synced with
#' their leaves order values.
#'
#' @param ... not used
#'
#' @details
#'
#' Untangaling two trees is a hard combinatorical problem without a closed
#' form solution. One way for doing it is to run through a random spectrom
#' of options and look for the "best" two trees. This is what this function
#' offers.
#'
#' @return A dendlist with two trees with the best entanglement that was found.
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}.
#' @examples
#'
#' \dontrun{
#' dend1 <- iris[, -5] %>%
#' dist() %>%
#' hclust("com") %>%
#' as.dendrogram()
#' dend2 <- iris[, -5] %>%
#' dist() %>%
#' hclust("sin") %>%
#' as.dendrogram()
#' tanglegram(dend1, dend2)
#'
#' set.seed(65168)
#' dend12 <- untangle_random_search(dend1, dend2, R = 10)
#' tanglegram(dend12[[1]], dend12[[2]])
#' tanglegram(dend12)
#'
#' entanglement(dend1, dend2, L = 2) # 0.8894
#' entanglement(dend12[[1]], dend12[[2]], L = 2) # 0.0998
#' }
untangle_random_search <- function(dend1, dend2, R = 100L, L = 1, leaves_matching_method = c("labels", "order"), ...) {
# this is a simple random search algorithm for the optimal tanglegram layout problem.
# it shufflers the trees, and see if we got a better entanglement or not
leaves_matching_method <- match.arg(leaves_matching_method)
if (leaves_matching_method == "order") {
old_dend2 <- dend2
dend2 <- match_order_by_labels(old_dend2, dend1)
if (!identical(dend2, old_dend2) & dendextend_options("warn")) warning("The leaves order in 'dend2' were changed. If you want to avoid that, use leaves_matching_method = 'labels'.")
}
optimal_dend1 <- dend1
optimal_dend2 <- dend2
best_ordaring_entanglement <- entanglement(dend1, dend2, L, leaves_matching_method)
for (i in 1:R) {
s_dend1 <- shuffle(dend1)
s_dend2 <- shuffle(dend2)
current_entanglement <- entanglement(s_dend1, s_dend2, L, leaves_matching_method)
# if we came across a better ordaring, then update the "Best" treerograms
if (current_entanglement < best_ordaring_entanglement) {
best_ordaring_entanglement <- current_entanglement
optimal_dend1 <- s_dend1
optimal_dend2 <- s_dend2
}
}
return(dendlist(optimal_dend1, optimal_dend2))
}
flip_strings <- function(STRING, str1, str2) {
# gets a string which includes str1 and str2, and makes sure to flip them in the string
STRING <- sub(str1, "_____1_", STRING, fixed = T) # substitutes the first string with a place holder (1)
STRING <- sub(str2, "_____2_", STRING, fixed = T) # substitutes the second string with a place holder (2)
STRING <- sub("_____1_", str2, STRING, fixed = T) # substitutes the place holder (1) with the second string
STRING <- sub("_____2_", str1, STRING, fixed = T) # substitutes the place holder (2) with the first string
return(STRING)
}
# flip_strings("abcdefgh", "ab", "fgh") # "fghcdeab"
add_zzz <- function(x) {
# this function adds a"_" character to the end of every element of the vector.
# this is used to make numeric values unique (so to not confuse 1 and 10 or 17 and 7 !)
x <- as.character(x)
x <- paste("zzz", x, "zzz", sep = "")
x
}
remove_zzz <- function(x) {
gsub("zzz", "", x, fixed = T)
}
# remove_zzz(add_zzz(1:6))
collapse_with_pipes <- function(x) {
paste(x, collapse = "||")
}
collapse_pipes_zzz <- function(x) {
paste(add_zzz(x), collapse = "||")
}
remove_pipes_and_zzz <- function(x) {
strsplit(remove_zzz(x), "||", fixed = T)[[1]]
}
#' @title Flip leaves
#' @export
#' @description
#' Rotate a branch in a tree so that the locations of two bundles of leaves
#' are flipped.
#'
#' @param dend a dendrogram object
#' @param leaves1 a vector of leaves order value to flip.
#' @param leaves2 a (second) vector of leaves order value to flip.
#' @param ... not used
#' @details
#' This function is based on a bunch of string manipulation functions. There
#' may be a smarter/better way for doing it...
#'
#' @return A dendrogram object with flipped leaves.
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}.
#' @examples
#'
#' \dontrun{
#' dend1 <- USArrests[1:5, ] %>%
#' dist() %>%
#' hclust() %>%
#' as.dendrogram()
#' dend2 <- flip_leaves(dend1, c(3, 5), c(1, 2))
#' tanglegram(dend1, dend2)
#' entanglement(dend1, dend2, L = 2) # 0.4
#' }
flip_leaves <- function(dend, leaves1, leaves2, ...) {
# flip a node in a tree based on the leaves in each branch in the node:
# this function gets a dendgram with two vector of leaves that needs to be flipped with one another on the tree
# we assume here unique values of leaves.
# also notice that this is based on the values of the leaves and NOT their labels.
leaves_order <- order.dendrogram(dend)
weights <- seq_along(leaves_order)
# turn the values of leaves and leaves1/2 to strings with || delim:
leaves_order_string <- collapse_pipes_zzz(leaves_order)
leaves1_string <- collapse_pipes_zzz(leaves1)
leaves2_string <- collapse_pipes_zzz(leaves2)
# then flips the locations of leaves1 and 2 in the string
flipped_leaves_order_string <- flip_strings(leaves_order_string, leaves1_string, leaves2_string)
# and turn the string back to a vector of flipped leaves values:
flipped_leaves_order <- as.integer(remove_pipes_and_zzz(flipped_leaves_order_string))
new_order_weights <- match(flipped_leaves_order, leaves_order) # order the leaves_order to be like flipped_leaves_order
# leaves_order[new_order_weights]
# now use this order to order the weights!
new_weights <- weights[new_order_weights]
flipped_dend <- rotate(dend, new_weights) # and lastly - rotate the dend by the leaves to flip.
return(flipped_dend)
}
# I didn't use this evantually:
# library(combinat)
# source for this package: https://stackoverflow.com/questions/7906332/how-to-calculate-combination-and-permutation-in-r
#' @title Rotate tree branches for k
#' @export
#' @description
#' Given a tree and a k number of clusters, the tree is rotated so that the
#' extra clusters added from k-1 to k clusters are flipped.
#'
#' This is useful for finding good trees for a \link{tanglegram}.
#' @param dend a dendrogram object
#' @param k integer scalar with the number of clusters the tree should be cut into.
#' @param dend_heights_per_k a named vector that resulted from running
#' \link{heights_per_k.dendrogram}. When running the function many times,
#' supplying this object will help improve the running time if using the
#' \link{cutree.dendrogram} method..
#'
#' @param ... not used
#' @return A list with dendrogram objects with all the possible rotations
#' for k clusters (beyond the k-1 clusters!).
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}, \link{flip_leaves}.
#' @examples
#'
#' \dontrun{
#' dend1 <- USArrests[1:5, ] %>%
#' dist() %>%
#' hclust() %>%
#' as.dendrogram()
#' dend2 <- all_couple_rotations_at_k(dend1, k = 2)[[2]]
#' tanglegram(dend1, dend2)
#' entanglement(dend1, dend2, L = 2) # 0.5
#'
#' dend2 <- all_couple_rotations_at_k(dend1, k = 3)[[2]]
#' tanglegram(dend1, dend2)
#' entanglement(dend1, dend2, L = 2) # 0.4
#'
#' dend2 <- all_couple_rotations_at_k(dend1, k = 4)[[2]]
#' tanglegram(dend1, dend2)
#' entanglement(dend1, dend2, L = 2) # 0.05
#' }
all_couple_rotations_at_k <- function(dend, k, dend_heights_per_k, ...) {
# This function gets the dend tree, and a k number of clusters
# and returns all of the permutated dendrogram trees, rotating only two of the k clusters at each permutation
# if this was done for ALL permutation, the algorithm would not be feasable.
# practically, for a binary tree - this only gives two trees as an output (the original, and the flipped new k'th cluster)
if (length(k) != 1L) {
warning("'k' should be an integer SCALAR, using only the first element of k.")
k <- k[1L]
}
if (k == 1) {
return(dend)
} # there are no possible rotations for k==1
if (missing(dend_heights_per_k)) {
dend_heights_per_k <- heights_per_k.dendrogram(dend)
} # since this function takes a looong time, I'm running it here so it will need to run only once!
# And I would MUCH rather give this vector upfront - so the entire thing will be faster...
leaves_order <- order.dendrogram(dend)
k_cluster_leaves <- cutree(dend, k,
order_clusters_as_data = FALSE,
dend_heights_per_k = dend_heights_per_k, # makes it faster
use_labels_not_values = FALSE
) # makes it 10 times faster (and we don't use the labels of the clusters, only the cluster vector)
km1_cluster_leaves <- cutree(dend, k - 1,
order_clusters_as_data = FALSE,
dend_heights_per_k = dend_heights_per_k, # makes it faster
use_labels_not_values = FALSE
) # makes it 10 times faster (and we don't use the labels of the clusters, only the cluster vector)
# if we can't cut the current stage (for example, because we have more than 2 branches, than return the original tree
if (any(is.na(k_cluster_leaves))) {
return(list(dend))
}
# If we can't cut the tree above us, then loop up until you find a k for which you can cut.
# there might be bugs for this code, more careful thought should be made in such cases...
while (any(is.na(km1_cluster_leaves))) {
k <- k - 1
km1_cluster_leaves <- cutree(dend, k - 1,
order_clusters_as_data = FALSE,
dend_heights_per_k = dend_heights_per_k, # makes it faster
use_labels_not_values = FALSE
) # makes it 10 times faster (and we don't use the labels of the clusters, only the cluster vector)
warning(paste("couldn't cut tree at k-1, trying it for", k - 1))
}
# kkm1_df <-
# data.frame(km1_cluster_leaves, k_cluster_leaves)
permutated_dend <- list(dend) # this one will hold all of the permutations
permutation_indx <- 1 # this one will tell us at what stage of the permutation we are at
for (i in unique(km1_cluster_leaves)) {
ss <- i == km1_cluster_leaves
unique_clusters_in_branch <- unique(k_cluster_leaves[ss])
if (length(unique_clusters_in_branch) > 1) { # the only way there is a reason to do permutations here is if the current cluster we are looking at has more than 1 member
number_of_clusters_in_branch <- length(unique_clusters_in_branch)
branches_permutations <- as.matrix(combn(unique_clusters_in_branch, 2)) # a matrix were each column is a permutation of 2 out of the clusters in this branch (most often just 2, but sometimes more...)
# as.matrix(combn(1:3, 2))
# permn(number_of_clusters_in_branch) # this will be 2 most of the time, but this structure allows one to deal with clusters which have more than 2 branches
n_permutations <- ncol(branches_permutations)
for (j in seq_len(n_permutations)) { # would often run just once.
# choosing the leaves belonging to each of the two clusters
ss_leaves1 <- k_cluster_leaves == branches_permutations[1, j]
ss_leaves2 <- k_cluster_leaves == branches_permutations[2, j]
leaves1 <- leaves_order[ss_leaves1]
leaves2 <- leaves_order[ss_leaves2]
# plot(flip_leaves(dend, leaves1, leaves2))
# Flipping the branches of the two adjecent clusters:
permutation_indx <- permutation_indx + 1
permutated_dend[[permutation_indx]] <- flip_leaves(dend, leaves1, leaves2) # this will not work for hclust (will for dend)
}
}
}
return(permutated_dend)
}
#' @title Stepwise untangle one tree compared to another
#' @export
#' @description
#' Given a fixed tree and a tree we wish to rotate, this function goes
#' through all of the k number of clusters (from 2 onward), and each time
#' rotates the branch which was introduced in the new k'th cluster.
#' This rotated tree is compared with the fixed tree, and if it has a better
#' entanglement, it will be used for the following iterations.
#'
#' This is a greedy forward selection algorithm for rotating the tree and
#' looking for a better match.
#'
#' This is useful for finding good trees for a \link{tanglegram}.
#' @param dend1 a dendrogram object. The one we will rotate to best fit
#' dend2_fixed.
#' @param dend2_fixed a dendrogram object. This one is kept fixed.
#' @param L the distance norm to use for measuring the distance between the
#' two trees. It can be any positive number,
#' often one will want to use 0, 1, 1.5, 2 (see 'details' in \link{entanglement}).
#'
#' @param direction a character scalar, either "forward" (default) or "backward".
#' Impacts the direction of clustering that are tried. Either from 2 and up
#' (in case of "forward"), or from nleaves to down (in case of "backward")
#'
#' If k_seq is not NULL, then it overrides "direction".
#'
#' @param k_seq a sequence of k clusters to go through for improving
#' dend1. If NULL (default), then we use the "direction" parameter.
#'
#' @param dend_heights_per_k a numeric vector of values which indicate which height will produce which number of clusters (k)
#'
#' @param leaves_matching_method a character scalar passed to \link{entanglement}.
#' It can be either "order" or "labels" (default). If using "labels",
#' then we use the labels for matching the leaves order value.
#' And if "order" then we use the old leaves order value for matching the
#' leaves order value.
#'
#' Using "order" is faster, but "labels" is safer. "order" will assume that
#' the original two trees had their labels and order values MATCHED.
#'
#' Hence, it is best to make sure that the trees used here have the same labels
#' and the SAME values matched to these values - and then use "order" (for
#' fastest results).
#'
#' If "order" is used, the function first calls \link{match_order_by_labels}
#' in order to make sure that the two trees have their labels synced with
#' their leaves order values.
#'
#' @param ... not used
#'
#' @return A dendlist with
#' 1) dend1 after it was rotated to best fit dend2_fixed.
#' 2) dend2_fixed.
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}, \link{flip_leaves}, \link{all_couple_rotations_at_k},
#' \link{untangle_step_rotate_2side}.
#'
#' @examples
#'
#' \dontrun{
#' dend1 <- USArrests[1:10, ] %>%
#' dist() %>%
#' hclust() %>%
#' as.dendrogram()
#' set.seed(3525)
#' dend2 <- shuffle(dend1)
#' tanglegram(dend1, dend2)
#' entanglement(dend1, dend2, L = 2) # 0.4727
#'
#' dend2_corrected <- untangle_step_rotate_1side(dend2, dend1)[[1]]
#' tanglegram(dend1, dend2_corrected) # FIXED.
#' entanglement(dend1, dend2_corrected, L = 2) # 0
#' }
untangle_step_rotate_1side <- function(dend1, dend2_fixed, L = 1.5, direction = c("forward", "backward"),
k_seq = NULL, dend_heights_per_k, leaves_matching_method = c("labels", "order"), ...) {
# this function gets two dendgrams, and goes over each k splits of the first dend1, and checks if the flip at level k of splitting imporves the entanglement between dend1 and dend2 (Which is fixed)
n_leaves <- nleaves(dend1)
best_dend <- dend1
if (missing(dend_heights_per_k)) dend_heights_per_k <- heights_per_k.dendrogram(best_dend) # since this function takes a looong time, I'm running it here so it will need to run only once!
leaves_matching_method <- match.arg(leaves_matching_method)
if (leaves_matching_method == "order") {
old_dend2_fixed <- dend2_fixed
dend2_fixed <- match_order_by_labels(old_dend2_fixed, dend1)
if (!identical(dend2_fixed, old_dend2_fixed) & dendextend_options("warn")) warning("The leaves order in 'dend2_fixed' were changed. If you want to avoid that, use leaves_matching_method = 'labels'.")
}
direction <- match.arg(direction)
if (is.null(k_seq)) {
# choose step direction:
if (direction == "backward") {
k_seq <- n_leaves:2
} else { # forward
k_seq <- 2:n_leaves
}
}
for (k in k_seq) {
dend1_k_rotated <- all_couple_rotations_at_k(best_dend, k, dend_heights_per_k = dend_heights_per_k)
dend1_cut_k_entanglements <- lapply(dend1_k_rotated, entanglement, dend2 = dend2_fixed, L = L, leaves_matching_method = leaves_matching_method)
ss_best_dend <- which.min(dend1_cut_k_entanglements)
current_best_dend <- dend1_k_rotated[[ss_best_dend]]
# if this loop's best dendro is not identical to our last best dendro - then we should pick it as the new best dendro
# And that means we'll have to update the heights_per_k.dendrogram (which takes time, and we would like to avoid if it is not necessary)
if (!identical(current_best_dend, best_dend)) {
best_dend <- current_best_dend
# We don't need to run the next line twice since the heights per k are the same for any rotated tree...
# best_dend_heights_per_k <- heights_per_k.dendrogram(best_dend)
} # however, if the current dend is just like our best dend - then there is NO NEED to update heights_per_k.dendrogram (and we just saved some time!!)
# this combination is only useful if we have a tree for which there are only a few rotations which are useful
}
return(dendlist(best_dend = best_dend, dend2_fixed = dend2_fixed))
}
#' @title Stepwise untangle two trees one at a time
#' @export
#' @description
#' This is a greedy forward selection algorithm for rotating the tree and
#' looking for a better match.
#'
#' This is useful for finding good trees for a \link{tanglegram}.
#'
#' It goes through rotating dend1, then dend2, and so on - until a locally optimal solution is found.
#'
#' Similar to "step1side", one tree is held fixed and the other tree is rotated.
#' This function goes through all of the k number of clusters (from 2 onward),
#' and each time rotates the branch which was introduced in the new k'th cluster.
#' This rotated tree is compared with the fixed tree, and if it has a better
#' entanglement, it will be used for the following iterations.
#' Once finished the rotated tree is held fixed, and the fixed tree
#' is now rotated. This continues until a local optimal solution is reached.
#'
#' @param dend1 a dendrogram object. The one we will rotate to best fit
#' dend2.
#' @param dend2 a dendrogram object. The one we will rotate to best fit
#' dend1.
#' @param L the distance norm to use for measuring the distance between the
#' two trees. It can be any positive number,
#' often one will want to use 0, 1, 1.5, 2 (see 'details' in \link{entanglement}).
#'
#' @param direction a character scalar, either "forward" (default) or "backward".
#' Impacts the direction of clustering that are tried. Either from 2 and up
#' (in case of "forward"), or from nleaves to down (in case of "backward")
#'
#' If k_seq is not NULL, then it overrides "direction".
#'
#' @param max_n_iterations integer. The maximal number of times to switch between optimizing one tree with another.
#' @param print_times logical (TRUE), should we print how many times we switched between rotating the two trees?
#' @param k_seq a sequence of k clusters to go through for improving
#' dend1. If NULL (default), then we use the "direction" parameter.
#' @param ... not used
#'
#' @return A list with two dendrograms (dend1/dend2),
#' after they are rotated to best fit one another.
#'
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}, \link{flip_leaves}, \link{all_couple_rotations_at_k}.
#' \link{untangle_step_rotate_1side}.
#' @examples
#'
#' \dontrun{
#' dend1 <- USArrests[1:20, ] %>%
#' dist() %>%
#' hclust() %>%
#' as.dendrogram()
#' dend2 <- USArrests[1:20, ] %>%
#' dist() %>%
#' hclust(method = "single") %>%
#' as.dendrogram()
#' set.seed(3525)
#' dend2 <- shuffle(dend2)
#' tanglegram(dend1, dend2, margin_inner = 6.5)
#' entanglement(dend1, dend2, L = 2) # 0.79
#'
#' dend2_corrected <- untangle_step_rotate_1side(dend2, dend1)
#' tanglegram(dend1, dend2_corrected, margin_inner = 6.5) # Good.
#' entanglement(dend1, dend2_corrected, L = 2) # 0.0067
#' # it is better, but not perfect. Can we improve it?
#'
#' dend12_corrected <- untangle_step_rotate_2side(dend1, dend2)
#' tanglegram(dend12_corrected[[1]], dend12_corrected[[2]], margin_inner = 6.5) # Better...
#' entanglement(dend12_corrected[[1]], dend12_corrected[[2]], L = 2) # 0.0045
#'
#'
#' # best combination:
#' dend12_corrected_1 <- untangle_random_search(dend1, dend2)
#' dend12_corrected_2 <- untangle_step_rotate_2side(dend12_corrected_1[[1]], dend12_corrected_1[[2]])
#' tanglegram(dend12_corrected_2[[1]], dend12_corrected_2[[2]], margin_inner = 6.5) # Better...
#' entanglement(dend12_corrected_2[[1]], dend12_corrected_2[[2]], L = 2) # 0 - PERFECT.
#' }
untangle_step_rotate_2side <- function(dend1, dend2, L = 1.5, direction = c("forward", "backward"), max_n_iterations = 10L, print_times = dendextend_options("warn"),
k_seq = NULL, ...) {
# this function gets two dendgrams, and orders dend1 and 2 until a best entengelment is reached.
direction <- match.arg(direction)
dend1_heights_per_k <- heights_per_k.dendrogram(dend1)
dend2_heights_per_k <- heights_per_k.dendrogram(dend2)
# Next, let's try to improve upon this tree using a forwared rotation of our tree:
dend1_better <- untangle_step_rotate_1side(dend1, dend2, L = L, dend_heights_per_k = dend1_heights_per_k, direction = direction, k_seq = k_seq)[[1]]
dend2_better <- untangle_step_rotate_1side(dend2, dend1_better, L = L, dend_heights_per_k = dend2_heights_per_k, direction = direction, k_seq = k_seq)[[1]]
entanglement_new <- entanglement(dend1_better, dend2_better, L = L)
entanglement_old <- entanglement_new + 1
times <- 1
while (times < max_n_iterations & !identical(entanglement_new, entanglement_old)) { # if we got an improvement from last entaglement, we'll keep going!
entanglement_old <- entanglement_new
dend1_better_loop <- untangle_step_rotate_1side(dend1_better, dend2_better,
L = L,
dend_heights_per_k = dend1_heights_per_k, direction = direction, k_seq = k_seq
)[[1]]
# if the new dend1 is just like we just had - then we can stop the function since we found the best solution - else - continue
if (identical(dend1_better_loop, dend1_better)) {
break
} else {
dend1_better <- dend1_better_loop
}
# if the new dend2 is just like we just had - then we can stop the function since we found the best solution - else - continue
dend2_better_loop <- untangle_step_rotate_1side(dend2_better, dend1_better,
L = L,
dend_heights_per_k = dend2_heights_per_k, direction = direction, k_seq = k_seq
)[[1]]
if (identical(dend2_better_loop, dend2_better)) {
break
} else {
dend2_better <- dend2_better_loop
}
entanglement_new <- entanglement(dend1_better, dend2_better, L = L)
times <- times + 1
}
# identical(1,1+.00000000000000000000000001) # T
if (print_times) cat("\nWe ran untangle ", times, " times\n")
return(dendlist(dend1_better, dend2_better))
}
#' @title Stepwise untangle two trees at the same time
#' @export
#' @description
#' This is a greedy forward selection algorithm for rotating the tree and
#' looking for a better match.
#'
#' This is useful for finding good trees for a \link{tanglegram}.
#'
#' It goes through simultaneously rotating branches of dend1 and dend2
#' until a locally optimal solution is found.
#'
#'
#' Step 1: The algorithm begins by executing the 'step2side' operation on the pair
#' of dendograms.
#'
#' Step 2: The algorithm generates new alternative tanglegrams by simultaneously
#' rotating one branch from tree 1 and one branch from tree 2. This rotation is
#' applied to every possible combination of branches between tree 1 and tree 2,
#' resulting in a set of new alternative tanglegrams. The tanglegram with the lowest
#' entanglement is retained.
#'
#' Step 3: Steps 1 and 2 are repeated until either a locally optimal solution is
#' found or the maximum number of iterations is reached.
#'
#' @param dend1 a dendrogram object. The one we will rotate to best fit
#' dend2.
#' @param dend2 a dendrogram object. The one we will rotate to best fit
#' dend1.
#' @param L the distance norm to use for measuring the distance between the
#' two trees. It can be any positive number,
#' often one will want to use 0, 1, 1.5, 2 (see 'details' in \link{entanglement}).
#'
#' @param max_n_iterations integer. The maximal number of times to switch between optimizing one tree with another.
#' @param print_times logical (TRUE), should we print how many times we executed steps 1 and 2?
#' @param ... not used
#'
#' @return A list with two dendrograms (dend1/dend2),
#' after they are rotated to best fit one another.
#'
#' @seealso \link{tanglegram}, \link{match_order_by_labels},
#' \link{entanglement}, \link{flip_leaves}, \link{all_couple_rotations_at_k}.
#' \link{untangle_step_rotate_1side}, \link{untangle_step_rotate_2side}.
#' @references
#' Nghia Nguyen, Kurdistan Chawshin, Carl Fredrik Berg, Damiano Varagnolo, Shuffle & untangle: novel untangle methods for solving the tanglegram layout problem, Bioinformatics Advances, Volume 2, Issue 1, 2022, vbac014, https://doi.org/10.1093/bioadv/vbac014
#'
#' @examples
#'
#' \dontrun{
#' # Figures recreated from 'Shuffle & untangle: novel untangle
#' # methods for solving the tanglegram layout problem' (Nguyen et al. 2022)
#' library(tidyverse)
#' example_labels <- c("Versicolor 90", "Versicolor 54", "Versicolor 81",
#' "Versicolor 63", "Versicolor 72", "Versicolor 99", "Virginica 135",
#' "Virginica 117", "Virginica 126", "Virginica 108", "Virginica 144",
#' "Setosa 27", "Setosa 18", "Setosa 36", "Setosa 45", "Setosa 9")
#'
#' iris_modified <-
#' iris %>%
#' mutate(Row = row_number()) %>%
#' mutate(Label = paste(str_to_title(Species), Row)) %>%
#' filter(Label %in% example_labels)
#' iris_numeric <- iris_modified[,1:4]
#' rownames(iris_numeric) <- iris_modified$Label
#'
#' # Single Linkage vs. Complete Linkage comparison (Fig. 1)
#' dend1 <- as.dendrogram(hclust(dist(iris_numeric), method = "single"))
#' dend2 <- as.dendrogram(hclust(dist(iris_numeric), method = "complete"))
#' tanglegram(dend1, dend2,
#' color_lines = TRUE,
#' lwd = 2,
#' margin_inner = 6) # Good.
#' entanglement(dend1, dend2, L = 2) # 0.207
#'
#' # The step2side algorithm (Fig. 2)
#' result <- untangle_step_rotate_2side(dend1, dend2)
#' tanglegram(result[[1]], result[[2]],
#' color_lines = TRUE,
#' lwd = 2,
#' margin_inner = 6) # Better...
#' entanglement(result[[1]], result[[2]], L = 2) # 0.185
#'
#' # The stepBothSides algorithm (Fig. 4)
#' result <- untangle_step_rotate_both_side(dend1, dend2)
#' tanglegram(result[[1]], result[[2]],
#' color_lines = TRUE,
#' lwd = 2,
#' margin_inner = 6,
#' lty = 1) # PERFECT.
#' entanglement(result[[1]], result[[2]], L = 2) # 0.000
#' }
untangle_step_rotate_both_side <- function(dend1, dend2, L = 1.5, max_n_iterations = 10L, print_times = dendextend_options("warn"), ...) {
# Implemented as described by pseudo-code in the paper 'Shuffle & untangle: novel untangle methods for solving the tanglegram layout problem' (Nguyen et al. 2022)
# Initialize placeholder values to be overwritten in first iteration
entanglement_new <- 0
entanglement_old <- 1
# Step 3: Repeat Steps 1 and 2 until the entanglement does not reduce any further
times <- 1
while (times < max_n_iterations & !identical(entanglement_new, entanglement_old)) {
# Step 1: Run the step2side algorithm until convergence
result <- untangle_step_rotate_2side(dend1, dend2, L = L, max_n_iterations = max_n_iterations, ...)
dend1 <- result[[1]]
dend2 <- result[[2]]
entanglement_old <- entanglement_new # Record best entanglement score from last iteration
entanglement_new <- entanglement(dend1, dend2, L = L)
# Step 2: Create new alternative tanglegrams by rotating both dendrograms simultaneously
n_leaves <- nleaves(dend1)
for (i in 1:(n_leaves - 1)) {
for (j in 1:(n_leaves - 1)) {
dend1_rotated <- suppressWarnings(rotate(dend1, i))
dend2_rotated <- suppressWarnings(rotate(dend2, j))
new_entanglement <- entanglement(dend1_rotated, dend2_rotated, L = L)
if (new_entanglement < entanglement_new) {
dend1 <- dend1_rotated
dend2 <- dend2_rotated
entanglement_new <- new_entanglement
}
}
}
times <- times + 1
}
if (print_times) cat("\nWe ran untangle ", times, " times\n")
return(dendlist(dend1, dend2))
}
##### Other attempts which have not
##### proven themselves as useful...
#
# untangle.forward.step.rotate.1side <- function(dend1, dend2_fixed) {
# # this function gets two dendgrams, and goes over each k splits of the first dend1, and checks if the flip at level k of splitting imporves the entanglement between dend1 and dend2 (Which is fixed)
# leaves_order <- order.dendrogram(dend1)
# best_dend <- dend1
#
# k_visited <- rep(F, length(leaves_order))
# k_visited[1] <- T # I don't need the first one
# k <- 1
#
# while(!all(k_visited)) {
# # create all of the rotations with k+-1:
# dend1_k_p1_rotated <- all_couple_rotations_at_k(best_dend, k+1)
# dend1_k_m1_rotated <- all_couple_rotations_at_k(best_dend, k-1)
# # find the enteglement for all of them:
# dend1_cut_k_p1_entanglements <- lapply(dend1_k_p1_rotated, entanglement, dend2 = dend2_fixed)
# dend1_cut_k_m1_entanglements <- lapply(dend1_k_m1_rotated, entanglement, dend2 = dend2_fixed)
# # what is best, forward or backward?
# if(min(dend1_cut_k_p1_entanglements) > min(dend1_cut_k_m1_entanglements)) {
#
# }
# k <- k + 1
# ss_best_dend <- which.min(dend1_cut_k_entanglements)
# best_dend <- dend1_k_rotated[[ss_best_dend]]
#
# all_couple_rotations_at_k(best_dend, -1)
# }
#
# return(best_dend)
# }
# dend12s_1_better <- untangle_step_rotate_1side(dend1, dend2)
# cutree(dend1, 10)
# evolution algorithm
untangle_intercourse <- function(brother_1_dend1, brother_1_dend2,
sister_2_dend1, sister_2_dend2, L = 1) {
# Gets two pairs of dend, and returns two childrens (inside a list)
children_1 <- untangle_step_rotate_2side(brother_1_dend1, brother_1_dend2, L = L)
children_2 <- untangle_step_rotate_2side(sister_2_dend1, sister_2_dend2, L = L)
dendlist(children_1, children_2)
}
entanglement_return_best_brother <- function(brother_1_dend1, brother_1_dend2,
brother_2_dend1, brother_2_dend2, L = 1) {
# Gets two pairs of dend, and returns the pair with the best (minimal) entanglement
if (entanglement(brother_1_dend1, brother_1_dend2, L = L) <
entanglement(brother_2_dend1, brother_2_dend2, L = L)) {
return(dendlist(brother_1_dend1, brother_1_dend2))
} else {
return(dendlist(brother_2_dend1, brother_2_dend2))
}
}
untangle_intercourse_evolution <- function(intercourse, L = 1) {
# intercourse is a list with two elements. Each element has two dends
entanglement_return_best_brother(intercourse[[1]], intercourse[[2]],
intercourse[[3]], intercourse[[4]],
L = L
)
}
untangle_evolution <- function(brother_1_dend1, brother_1_dend2,
sister_2_dend1, sister_2_dend2, L = 1) {
intercourse <- untangle_intercourse(brother_1_dend1, brother_1_dend2,
sister_2_dend1, sister_2_dend2,
L = L
) # creates a list with two pairs of dends
untangle_intercourse_evolution(intercourse, L = L) # returns the best child
}
####
# A new approuch - I will go through every possible flip on one side, and find the one that gives the best improvement.
# I will do the same on each tree, back and forth, until no better flip is found.
untangle_best_k_to_rotate_by_1side <- function(dend1, dend2_fixed, L = 1) {
# this function gets two dendgrams, and goes over each k splits of the first dend1, and checks if the flip at level k of splitting imporves the entanglement between dend1 and dend2 (Which is fixed)
leaves_order <- order.dendrogram(dend1)
best_dend <- dend1
dend1_k_rotated <- NULL
best_dend_heights_per_k <- heights_per_k.dendrogram(best_dend) # since this function takes a looong time, I'm running it here so it will need to run only once!
# this makes the function about twice as fast...
for (k in 2:length(leaves_order)) {
dend1_k_rotated <- c(
dend1_k_rotated,
all_couple_rotations_at_k(best_dend, k,
dend_heights_per_k = best_dend_heights_per_k
)
)
}
dend1_cut_k_entanglements <- lapply(dend1_k_rotated, entanglement, dend2 = dend2_fixed, L = L)
ss_best_dend <- which.min(dend1_cut_k_entanglements)
best_dend <- dend1_k_rotated[[ss_best_dend]]
return(best_dend)
}
flip_1_and_2 <- function(x) {
ifelse(x == 1, 2, 1)
}
untangle_best_k_to_rotate_by_2side_backNforth <- function(dend1, dend2, times_to_stop = 2, print_times = T, L = 1) {
# this function gets two dendgrams, and orders dend1 and then 2 and then 1 again - back and forth -until a best entengelment is reached.
was_improved <- T # e.g: we can improve it further
counter <- 1
while (was_improved) {
entanglement_old <- entanglement(dend1, dend2, L = L)
dend1 <- untangle_best_k_to_rotate_by_1side(dend1, dend2, L = L)
dend2 <- untangle_best_k_to_rotate_by_1side(dend2, dend1, L = L)
entanglement_new <- entanglement(dend1, dend2, L = L)
was_improved <- identical(entanglement_old, entanglement_new)
counter <- counter + 1
}
# identical(1,1+.00000000000000000000000001) # T
if (print_times) cat("We ran untangle_best_k_to_rotate_by_2side_backNforth ", counter, " times")
return(dendlist(dend1, dend2))
}
#
#
# untangle_OLO <- function(dend1, dend2, ...) {
#
# if(is.dendlist(dend1)) {
# dend2 <- dend1[[2]]
# dend1 <- dend1[[1]]
# }
#
# # hmap(sqrt(d2), Colv = "none", trace = "none", col = viridis(200))
# # Error in (function (x, Rowv = TRUE, Colv = if (symm) "Rowv" else TRUE, :
# # formal argument "Colv" matched by multiple actual arguments
# # d <- cophenetic(dend2) # doesn't work so great
#
# vec <- cbind(order.dendrogram(dend1), order.dendrogram(dend2))
# rownames(vec) <- labels(dend1)[order.dendrogram(dend1)]
# d <- dist(vec)
# o <- seriate(d, method = "OLO", control = list(hclust = as.hclust(dend1)) )
# dend1 <- rotate(dend1, order = rev(labels(d)[get_order(o)]))
# # library(dendextend)
# # o <- seriate(d, method = "OLO", control = list(hclust = as.hclust(dend2)) )
# # dend2 <- rotate(dend2, order = labels(d)[get_order(o)])
# return(dendlist(dend1, dend2))
# }
#
#
#
# if(F) {
# ## Not run:
# require(dendextend)
# set.seed(23235)
# ss <- sample(1:150, 10 )
# dend1 <- iris[ss,-5] %>% dist %>% hclust("com") %>% as.dendrogram
# dend2 <- iris[ss,-5] %>% dist %>% hclust("sin") %>% as.dendrogram
# dend12 <- dendlist(dend1, sort(dend2, type = "nodes", decreasing= T))
# # dend12 <- dendlist(dend1, sort(dend1))
# dend12 %>% tanglegram
# dend12_OLO <- untangle_OLO(dend12)
# dend12_OLO %>% tanglegram
# dend12_OLO %>% sort(type = "nodes") %>% tanglegram
#
# }
#
# if(F) {
# # example
# dist_DATA <- dist(USArrests[1:20,])
# # First two dummy clusters (since you didn't provide with some...)
# hc1 <- hclust(dist_DATA , "single")
# hc2 <- hclust(dist_DATA , "complete")
# dend1 <- as.dendrogram(hc1)
# dend2 <- as.dendrogram(hc2)
# entanglement(dend1, dend2)
#
# system.time(dend12_best_01 <- untangle_step_rotate_2side(dend1, dend2, L = 2)) # 0.47 sec
# system.time(dend12_best_02 <- untangle_best_k_to_rotate_by_2side_backNforth(dend1, dend2, L = 2)) # 0.44 sec
# tanglegram(dend1, dend2)
# tanglegram(dend12_best_01[[1]], dend12_best_01[[2]])
# tanglegram(dend12_best_02[[1]], dend12_best_02[[2]])
# }
#
#
#
#
#
#
#
#
# richrach <- function(x) {
# # move back and forth between the beginning and the end of a vector
# c(t(cbind(x, rev(x))))[1:length(x)]
# # example:
# # richrach(1:6)
# # from this: 1 2 3 4 5 6
# # to this: 1 6 2 5 3 4
# }
#
# richrach_xy <- function(x,y) {
# # move back and forth between the beginning and the end of a vector
# c(t(cbind(x, y)))[1:length(x)]
# # example:
# # richrach(1:6)
# # from this: 1 2 3 4 5 6
# # to this: 1 6 2 5 3 4
# }
#
#
# odd_locations <- function(x) {
# x[seq(1, length(x), by = 2)]
# }
# # odd_locations(1:6)
#
# #
# #
# # if(FALSE) {
# #
# # dist_DATA <- dist(USArrests[1:30,])
# # dist_DATA <- dist(USArrests[1:10,])
# # # First two dummy clusters (since you didn't provide with some...)
# # hc1 <- hclust(dist_DATA , "single")
# # hc2 <- hclust(dist_DATA , "complete")
# # dend1 <- as.dendrogram(hc1)
# # dend2 <- as.dendrogram(hc2)
# #
# # tanglegram(dend1, dend2)
# # entanglement(dend1, dend2) # 0.8
# #
# # # after sorting we get a better measure of entanglement and also a better looking plot
# # tanglegram(sort(dend1), sort(dend2))
# # entanglement(sort(dend1), sort(dend2)) # 0.1818
# #
# # # let's cause some shuffle... (e.g: mix the dendrogram, and see how that effects the outcome)
# # set.seed(134)
# # s_dend1 <- shuffle(dend1)
# # s_dend2 <- shuffle(dend2)
# # tanglegram(s_dend1, s_dend2)
# # entanglement(s_dend1, s_dend2) # 0.7515
# #
# #
# # set.seed(1234)
# # dend12s <- untangle.random.search(dend1, dend2, R = 10)
# # entanglement(dend12s[[1]], dend12s[[2]]) # 0.042
# # tanglegram(dend12s[[1]], dend12s[[2]]) #
# # # this is a case where it is CLEAR that the simplest heuristic would improve this to 0 entanglement...
# #
# # # let's see if we can reach a good solution using a greedy forward selection algorithm
# # dend12s_1_better <- untangle_step_rotate_1side(dend12s[[1]], dend12s[[2]])
# # entanglement(dend12s_1_better, dend12s[[2]]) # from 0.042 to 0.006 !!
# # tanglegram(dend12s_1_better, dend12s[[2]]) #
# #
# # # let's see from the beginning
# # entanglement(dend1, dend2) # 0.6
# # tanglegram(dend1, dend2) # 0.6
# # dend12s_1_better <- untangle_step_rotate_1side(dend1, dend2)
# # entanglement(dend12s_1_better, dend2) # from 0.6 to 0.036
# # tanglegram(dend12s_1_better, dend2) #
# # # let's try the other side:
# # dend12s_2_better <- untangle_step_rotate_1side(dend2, dend12s_1_better)
# # entanglement(dend12s_1_better, dend12s_2_better) # no improvment
# #
# #
# #
# # dend2_01 <- untangle_step_rotate_1side(dend2, dend1)
# # dend2_02 <- untangle.backward.rotate.1side(dend2, dend1)
# # dend2_03 <- untangle.backward.rotate.1side(dend2_01, dend1)
# # dend2_04 <- untangle_step_rotate_1side(dend2_02, dend1)
# # dend2_05 <- untangle_evolution(dend1, dend2 , dend1, dend2_01 )
# # entanglement(dend1, dend2)
# # entanglement(dend1, dend2_01)
# #
# # entanglement(dend1, dend2_02)
# # entanglement(dend1, dend2_03)
# # entanglement(dend1, dend2_04)
# # entanglement(dend2_05[[1]], dend2_05[[2]])
# # tanglegram(dend1, dend2)
# # tanglegram(dend1, dend2_01)
# # tanglegram(dend1, dend2_02)
# # tanglegram(dend1, dend2_03)
# # tanglegram(dend1, dend2_04)
# # tanglegram(dend2_05[[1]], dend2_05[[2]])
# #
# #
# #
# # entanglement(dend1, dend2)
# # tanglegram(dend1, dend2)
# # dend2_01 <- untangle_step_rotate_1side(dend2, dend1)
# # dend2_01 <- untangle.backward.rotate.1side(dend2, dend1)
# # tanglegram(dend1, dend2_01)
# #
# #
# #
# # #
# # dist_DATA <- dist(USArrests[1:10,])
# # # First two dummy clusters (since you didn't provide with some...)
# # hc1 <- hclust(dist_DATA , "single")
# # hc2 <- hclust(dist_DATA , "complete")
# # dend1 <- as.dendrogram(hc1)
# # dend2 <- as.dendrogram(hc2)
# # dend1_01 <- untangle_step_rotate_1side(dend1, dend2)
# # entanglement(dend1, dend2)
# # entanglement(dend1_01, dend2)
# # tanglegram(dend1, dend2)
# # tanglegram(dend1_01, dend2)
# #
# # system.time(dend1_01 <- untangle_step_rotate_1side(dend1, dend2)) # 0.47 sec
# # system.time(dend1_01 <- untangle.best.k.to.rotate.by(dend1, dend2)) # 0.44 sec
# # tanglegram(dend1, dend2)
# # tanglegram(dend1_01, dend2)
# #
# #
# #
# #
# # #### profiling
# # library(profr)
# # slow_dude <- profr(untangle_step_rotate_1side(dend2, dend1))
# # head(slow_dude)
# # summary(slow_dude)
# # plot(slow_dude)
# #
# # library(reshape)
# # a <- cast(slow_dude, f~., value="time", fun.aggregate=c(length, sum))
# # a[order(a[,3]),]
# # ## End(Not run)
# # slow_dude[slow_dude$time > .079991, ]
# #
# #
# # # this also helped:
# # # install.packages("microbenchmark")
# # library(microbenchmark)
# #
# # system.time(entanglement(dend1, dend2) ) # 0.01
# # microbenchmark( entanglement(dend1, dend2) , times = 10 )# so this is 10 times faster (the medians)
# # # betweem 0.011 to 0.038
# #
# # }
# #
# #
# #
# #
# #
# #
# # if(FALSE){
# #
# # # Finding the BEST tree by going through many random seeds and looking for a good solution :)
# #
# # entanglement_history <- c()
# #
# #
# # get.seed <- function(max_lengh = 10e7) round(runif(1)*max_lengh)
# #
# # best_seed <- 28754448 # 55639690 # 5462457 # 75173309 # 20295644
# # set.seed(best_seed)
# # times_a_better_seed_was_found <- 0
# # random_dendros <- untangle.random.search(yoavs_tree, Dan_arc_tree, R = 1, L = 1.5)
# # rotated_dendros <- untangle_step_rotate_2side(random_dendros[[1]], random_dendros[[2]], L = 1.5)
# # best_entanglement <- entanglement(rotated_dendros[[1]], rotated_dendros[[2]], L = 1.5)
# # # tanglegram(rotated_dendros[[1]], rotated_dendros[[2]])
# #
# #
# # for(i in 1:100000) {
# # print(i)
# # current_seed <- get.seed()
# # set.seed(current_seed)
# # random_dendros <- untangle.random.search(yoavs_tree, Dan_arc_tree, R = 10, L = 1.5)
# # rotated_dendros <- untangle_step_rotate_2side(random_dendros[[1]], random_dendros[[2]], L = 1.5)
# # new_entanglement <- entanglement(rotated_dendros[[1]], rotated_dendros[[2]], L = 1.5)
# #
# # entanglement_history <- c(entanglement_history, new_entanglement)
# #
# # if(new_entanglement < best_entanglement ){
# # times_a_better_seed_was_found <- times_a_better_seed_was_found + 1
# # best_seed <- current_seed
# # print(best_seed)
# # print(new_entanglement)
# # best_entanglement <- new_entanglement
# # tanglegram(rotated_dendros[[1]], rotated_dendros[[2]])
# # }
# # flush.console()
# # }
# #
# #
# # hist(entanglement_history)
# #
# # }
# #
# #
# #
# #
# #
#
#
#
#
#
#
# # this function is from the combinat package
# # permn <- function (x, fun = NULL, ...)
# # {
# # if (is.numeric(x) && length(x) == 1 && x > 0 && trunc(x) ==
# # x)
# # x <- seq(x)
# # n <- length(x)
# # nofun <- is.null(fun)
# # out <- vector("list", gamma(n + 1))
# # p <- ip <- seqn <- 1:n
# # d <- rep(-1, n)
# # d[1] <- 0
# # m <- n + 1
# # p <- c(m, p, m)
# # i <- 1
# # use <- -c(1, n + 2)
# # while (m != 1) {
# # out[[i]] <- if (nofun)
# # x[p[use]]
# # else fun(x[p[use]], ...)
# # i <- i + 1
# # m <- n
# # chk <- (p[ip + d + 1] > seqn)
# # m <- max(seqn[!chk])
# # if (m < n)
# # d[(m + 1):n] <- -d[(m + 1):n]
# # index1 <- ip[m] + 1
# # index2 <- p[index1] <- p[index1 + d[m]]
# # p[index1 + d[m]] <- m
# # tmp <- ip[index2]
# # ip[index2] <- ip[m]
# # ip[m] <- tmp
# # }
# # out
# # }
# #
#
#
# #
# #
# #
# # order.weights.by.cluster.order <- function(weights, cluster_id, new_cluster_order) {
# # # this function gets a vector of weights. The clusters each weight belongs to
# # # and a new order for the clusters
# # # and outputs the new weight vector after ordering the vector by the new order of the cluster (internal order of elements within each cluster is preserved)
# # output <- NULL
# # for(i in seq_along(new_cluster_order)) {
# # output <- c(output, weights[cluster_id == new_cluster_order[i]])
# # }
# # return(output)
# # }
# # if(F){
# # # example:
# # x = c(1,2,3,4,5,6)
# # ord1 = c(1,1,2,2,2,3)
# # ord_of_clusters = c(2,1,3)
# # c(x[ord1 == ord_of_clusters[1]],x[ord1 == ord_of_clusters[2]],x[ord1 == ord_of_clusters[3]])
# # order.weights.by.cluster.order(x, ord1, ord_of_clusters)
# # }
#
#
#
#
#
#
|