Problem 300: Protein folding

(see projecteuler.net/problem=300)

In a very simplified form, we can consider proteins as strings consisting of hydrophobic (H) and polar (P) elements, e.g. HHPPHHHPHHPH.
For this problem, the orientation of a protein is important; e.g. HPP is considered distinct from PPH. Thus, there are 2n distinct proteins consisting of n elements.

When one encounters these strings in nature, they are always folded in such a way that the number of H-H contact points is as large as possible, since this is energetically advantageous.
As a result, the H-elements tend to accumulate in the inner part, with the P-elements on the outside.
Natural proteins are folded in three dimensions of course, but we will only consider protein folding in two dimensions.

The figure below shows two possible ways that our example protein could be folded (H-H contact points are shown with red dots).
protein folding
The folding on the left has only six H-H contact points, thus it would never occur naturally.
On the other hand, the folding on the right has nine H-H contact points, which is optimal for this string.

Assuming that H and P elements are equally likely to occur in any position along the string, the average number of H-H contact points in an optimal folding of a random protein string of length 8 turns out to be 850 / 28=3.3203125.

What is the average number of H-H contact points in an optimal folding of a random protein string of length 15?
Give your answer using as many decimal places as necessary for an exact result.

My Algorithm

The core idea is:
- generate all possible layouts (see search())
- for each layout: "fill" it with all possible proteins and count contact points (see optimize)

search() uses a 30x30 grid where the first element is placed at (15,15).
Since all rotated and mirrored proteins yield the same number of contact points I also place the second element at (16,15).
In each subsequent step all possible continuations are tried until 15 elements are placed on the grid.
I decided to label those elements a, b, ..., o (equivalent to First+0, First+1, ..., First+14). Empty cells are with a zero.

When 15 elements are stored in grid, then the function optimize finds all connections between cells.
E.g. in the example image there is a contact between 0 and 14 (the first and the last element of the protein).
It deliberately ignores immediate neighbors, such as 0 and 1 because they always have a contact independent of the protein's layout.
This accelerates the analysis because those contacts of immediate neighbors can be pre-computed efficiently in main() and reduce the time spent in optimize considerably.

Several proteins share the same layout and thus have the same number of contact points.
Even more, several protein are not identical but "similar": their layout varies but they have contact points at the same locations.
My "cache" named visited detects those identical and those similar proteins to speed up the process.

The contact points are based on bit masks: if element 3 and element 7 have a contact, then the bitmask will be (1 << 3) | (1 << 7).
When I iterate over all proteins then there will be 1 << length combinations. I decided that a zero bit refers to a hydrophobic element.
Thus only if (protein & contactMask) == 0 then both elements are hydrophobic.

Note

The grid is 30x30 because when I start in the middle of the grid then a protein which is a straight line could extend to (29,15).
Obviously this layout has only contact points between immediate neighboring elements and is not an optimal choice for most proteins.
On the other hand, proteins with many contact points are often very compact.
If I don't search the whole grid for potential contact points then I could miss a few.
As it turns out, missing contacts doesn't matter for sub-optimal layout - as long as I correctly process all optimal layouts.
And it's even okay to get a few optimal layouts wrong - as long as there as multiple optimal layout and I don't mess up all of them.
I played around with the value ignoreStraight: reducing the search window to the area (9,9) - (21,21) still produces the correct result.
This is only true for proteins with 15 elements, therefore I skip this optimization for other protein lengths.

Placing the second element always at (16,15) is possible because of the massive amount of rotated and mirrored proteins.
A similar optimization is to call optimize only if the last element's y >= Center (which is 15):
For each protein which the last element at (x,y) there will be a mirrored protein ended at (x, 2*Center - y).

I tried to implement another optimization:
for each protein with a sequence abc..mno there is a flipped protein onm...bca. Thus I could skip about half of all protein sequences.
In the end I couldn't achieve a speed-up and removed the code.

I liked this problem !

Interactive test

You can submit your own input to my program and it will be instantly processed at my server:

Input data (separated by spaces or newlines):

This is equivalent to
echo 8 | ./300

Output:

(please click 'Go !')

Note: the original problem's input 15 cannot be entered
because just copying results is a soft skill reserved for idiots.

(this interactive test is still under development, computations will be aborted after one second)

My code

… was written in C++11 and can be compiled with G++, Clang++, Visual C++. You can download it, too.

#include <iostream>
#include <iomanip>
#include <vector>
#include <set>
 
// operate on a 30x30 grid, each protein starts at (Center, Center)
const unsigned int GridSize = 30;
const unsigned int Center = GridSize / 2;
typedef char Grid[GridSize][GridSize];
 
// indicator for an empty cell
const char Empty = 0;
// protein's first element is 'a', then 'b', ..., 'e'
const char First = 'a';
 
// optimized number of hydrophobic contacts per protein
std::vector<unsigned int> best;
// number of direct contacts between immediate neighbors (minimum value, independent of folding)
std::vector<unsigned int> direct;
// keep track of past contact points sequences (detect duplicates / strong similiarities)
std::set<std::vector<unsigned int>> visited;
 
// check each protein whether the current grid is better than anything seen before
void optimize(unsigned int length, const Grid& grid)
{
// list of neighbors, each pair is stored as a bitmask
std::vector<unsigned int> contacts;
 
// count how many occupied cells I've seen
unsigned int seen = 0;
 
// extremely straight foldings can't be optimal because they have only contacts between directs neighbors
unsigned int ignoreStraight = 9; // empirical value
if (length != 15)
ignoreStraight = 0;
 
// walk through the grid
for (unsigned int i = ignoreStraight; i + 1 + ignoreStraight < GridSize; i++)
for (unsigned int j = ignoreStraight; j + 1 + ignoreStraight < GridSize; j++)
{
// already visited the entire protein ?
if (seen + 1 >= length)
break;
 
// ignore empty cells
if (grid[i][j] == Empty)
continue;
 
// one more element visited
seen++;
// get its index (0..length-1)
auto from = grid[i][j] - First;
 
// check right neighbor
if (grid[i + 1][j] != Empty)
{
auto to = grid[i + 1][j] - First;
if (from != to + 1 && from + 1 != to)
contacts.push_back((1 << from) | (1 << to));
}
// check bottom neighbor
if (grid[i][j + 1] != Empty)
{
auto to = grid[i][j + 1] - First;
if (from != to + 1 && from + 1 != to)
contacts.push_back((1 << from) | (1 << to));
}
}
 
// no contacts at all ? (besides direct contact)
if (contacts.empty())
return;
 
// already had a similar folding ? (same contact points)
if (visited.count(contacts) != 0)
return;
visited.insert(contacts);
 
// for each protein check whether more hydrophobic contacts exist
auto numProteins = 1 << length;
for (auto protein = 0; protein < numProteins; protein++)
{
// at least all direct contacts
auto found = direct[protein];
 
// can't beat the previous record ?
if (found + contacts.size() <= best[protein])
continue;
 
// count other contact points
for (auto contactMask : contacts)
{
// are both neighbors hydrophobic ?
if ((protein & contactMask) == 0)
found++;
}
 
// better than before ?
if (best[protein] < found)
best[protein] = found;
}
}
 
// generate recursive all possible layouts
void search(unsigned int current, unsigned int length, Grid& grid, unsigned int x, unsigned int y)
{
// protein completely placed on grid ?
if (current == length)
{
// don't check layouts mirrored along y-axis
if (y >= Center)
optimize(length, grid);
 
// done
return;
}
 
// try all possible continuations
// go left
if (grid[x - 1][y] == Empty)
{
grid[x - 1][y] = current + First;
search(current + 1, length, grid, x - 1, y);
grid[x - 1][y] = Empty;
}
 
// go right
if (grid[x + 1][y] == Empty)
{
grid[x + 1][y] = current + First;
search(current + 1, length, grid, x + 1, y);
grid[x + 1][y] = Empty;
}
 
// go down
if (grid[x][y - 1] == Empty)
{
grid[x][y - 1] = current + First;
search(current + 1, length, grid, x, y - 1);
grid[x][y - 1] = Empty;
}
 
// go up
if (grid[x][y + 1] == Empty)
{
grid[x][y + 1] = current + First;
search(current + 1, length, grid, x, y + 1);
grid[x][y + 1] = Empty;
}
}
 
int main()
{
unsigned int length = 15;
std::cin >> length;
 
// allocate memory
auto numProteins = 1 << length;
direct.resize(numProteins, 0);
best .resize(numProteins, 0);
 
// count direct neighboring hydrophobic elements in each protein
for (auto protein = 0; protein < numProteins; protein++)
{
for (unsigned int i = 0; i < length - 1; i++)
if ((protein & (3 << i)) == 0) // 3 => bitmask of two bits
direct[protein]++;
 
best[protein] = direct[protein];
}
 
// create an empty grid
Grid grid;
for (unsigned int x = 0; x < GridSize; x++)
for (unsigned int y = 0; y < GridSize; y++)
grid[x][y] = Empty;
 
// always start in the middle
unsigned int x = Center;
unsigned int y = Center;
grid[x][y] = First;
// due to symmetries I can also place the second element
x++;
grid[x][y] = First + 1;
 
// generate all layouts and then call optimize()
search(2, length, grid, x, y);
 
// count optimized contact points
unsigned int sum = 0;
for (auto x : best)
sum += x;
// display average
std::cout << std::setprecision(14)
<< sum / double(best.size()) << std::endl;
return 0;
}

This solution contains 30 empty lines, 42 comments and 4 preprocessor commands.

Benchmark

The correct solution to the original Project Euler problem was found in 2.6 seconds on a Intel® Core™ i7-2600K CPU @ 3.40GHz.
Peak memory usage was about 5 MByte.

(compiled for x86_64 / Linux, GCC flags: -O3 -march=native -fno-exceptions -fno-rtti -std=c++11 -DORIGINAL)

See here for a comparison of all solutions.

Note: interactive tests run on a weaker (=slower) computer. Some interactive tests are compiled without -DORIGINAL.

Changelog

August 27, 2017 submitted solution
August 27, 2017 added comments

Difficulty

50% Project Euler ranks this problem at 50% (out of 100%).

Heatmap

Please click on a problem's number to open my solution to that problem:

green   solutions solve the original Project Euler problem and have a perfect score of 100% at Hackerrank, too
yellow solutions score less than 100% at Hackerrank (but still solve the original problem easily)
gray problems are already solved but I haven't published my solution yet
blue solutions are relevant for Project Euler only: there wasn't a Hackerrank version of it (at the time I solved it) or it differed too much
orange problems are solved but exceed the time limit of one minute or the memory limit of 256 MByte
red problems are not solved yet but I wrote a simulation to approximate the result or verified at least the given example - usually I sketched a few ideas, too

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
The 270 solved problems (level 10) had an average difficulty of 31.3% at Project Euler and
I scored 13,386 points (out of 15600 possible points, top rank was 17 out of ≈60000 in August 2017) at Hackerrank's Project Euler+.

My username at Project Euler is stephanbrumme while it's stbrumme at Hackerrank.

Look at my progress and performance pages to get more details.

more about me can be found on my homepage, especially in my coding blog.
some names mentioned on this site may be trademarks of their respective owners.
thanks to the KaTeX team for their great typesetting library !