Mercurial > repos > rliterman > csp2

annotate CSP2/CSP2_env/env-d9b9114564458d9d-741b3de822f2aaca6c6caa4325c4afce/opt/bbmap-39.01-1/current/aligner/SingleStateAlignerFlat2Amino.java @ 68:5028fdace37b

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 2e9511a184a1ca667c7be0c6321a36dc4e3d116d
author jpayne
date Tue, 18 Mar 2025 16:23:26 -0400
parents
children
rev   line source
jpayne@68 1 package aligner;
jpayne@68 2
jpayne@68 3 import dna.AminoAcid;
jpayne@68 4 import shared.KillSwitch;
jpayne@68 5 import shared.Tools;
jpayne@68 6
jpayne@68 7 /**
jpayne@68 8 * Based on SSAFlat, but with previous state pointers removed. */
jpayne@68 9 public final class SingleStateAlignerFlat2Amino implements Aligner {
jpayne@68 10
jpayne@68 11
jpayne@68 12 public SingleStateAlignerFlat2Amino(){}
jpayne@68 13
jpayne@68 14 private void prefillTopRow(){
jpayne@68 15 final int[] header=packed[0];
jpayne@68 16 final int qlen=rows;
jpayne@68 17 for(int i=0; i<=columns; i++){
jpayne@68 18 int x=columns-i+1;
jpayne@68 19 int qbases=qlen-x;
jpayne@68 20
jpayne@68 21 //Minimal points to prefer a leftmost alignment
jpayne@68 22 header[i]=qbases<=0 ? 0 : -qbases;
jpayne@68 23
jpayne@68 24 //Forces consumption of query, but does not allow for insertions...
jpayne@68 25 // header[i]=qbases<=0 ? 0 : calcDelScoreOffset(qbases);
jpayne@68 26 }
jpayne@68 27 }
jpayne@68 28
jpayne@68 29 private void prefillLeftColumnStartingAt(int i){
jpayne@68 30 packed[0][0]=MODE_MATCH;
jpayne@68 31 i=Tools.max(1, i);
jpayne@68 32 for(int score=MODE_INS+(POINTS_INS*i); i<=maxRows; i++){//Fill column 0 with insertions
jpayne@68 33 score+=POINTS_INS;
jpayne@68 34 packed[i][0]=score;
jpayne@68 35 }
jpayne@68 36 }
jpayne@68 37
jpayne@68 38 private void initialize(int rows_, int columns_){
jpayne@68 39 rows=rows_;
jpayne@68 40 columns=columns_;
jpayne@68 41 if(rows<=maxRows && columns<=maxColumns){
jpayne@68 42 prefillTopRow();
jpayne@68 43 // prefillLeftColumn();
jpayne@68 44 return;
jpayne@68 45 }
jpayne@68 46
jpayne@68 47 final int maxRows0=maxRows;
jpayne@68 48 final int maxColumns0=maxColumns;
jpayne@68 49 final int[][] packed0=packed;
jpayne@68 50
jpayne@68 51 //Monotonic increase
jpayne@68 52 maxRows=Tools.max(maxRows, rows+10);
jpayne@68 53 maxColumns=Tools.max(maxColumns, columns+10);
jpayne@68 54
jpayne@68 55 if(packed==null || maxColumns>maxColumns0){//Make a new matrix
jpayne@68 56 packed=KillSwitch.allocInt2D(maxRows+1, maxColumns+1);
jpayne@68 57 prefillLeftColumnStartingAt(1);
jpayne@68 58 }else{//Copy old rows
jpayne@68 59 assert(maxRows0>0 && maxColumns0>0);
jpayne@68 60 assert(maxRows>maxRows0 && maxColumns<=maxColumns0);
jpayne@68 61 packed=KillSwitch.allocInt2D(maxRows+1);
jpayne@68 62 for(int i=0; i<packed.length; i++){
jpayne@68 63 if(i<packed0.length){
jpayne@68 64 packed[i]=packed0[i];
jpayne@68 65 }else{
jpayne@68 66 packed[i]=KillSwitch.allocInt1D(maxColumns+1);
jpayne@68 67 }
jpayne@68 68 }
jpayne@68 69 //Fill column 0 with insertions
jpayne@68 70 prefillLeftColumnStartingAt(maxRows0);
jpayne@68 71 }
jpayne@68 72 prefillTopRow();
jpayne@68 73 }
jpayne@68 74
jpayne@68 75 /** return new int[] {rows, maxCol, maxState, maxScore, maxStart};
jpayne@68 76 * Will not fill areas that cannot match minScore */
jpayne@68 77 @Override
jpayne@68 78 public final int[] fillLimited(byte[] read, byte[] ref, int refStartLoc, int refEndLoc, int minScore){
jpayne@68 79 return fillUnlimited(read, ref, refStartLoc, refEndLoc, minScore);
jpayne@68 80 }
jpayne@68 81
jpayne@68 82 @Override
jpayne@68 83 public final int[] fillUnlimited(byte[] read, byte[] ref, int refStartLoc, int refEndLoc){
jpayne@68 84 return fillUnlimited(read, ref, refStartLoc, refEndLoc, -999999);
jpayne@68 85 }
jpayne@68 86
jpayne@68 87 /** return new int[] {rows, maxCol, maxState, maxScore, maxStart};
jpayne@68 88 * Min score is optional */
jpayne@68 89 @Override
jpayne@68 90 public final int[] fillUnlimited(byte[] read, byte[] ref, int refStartLoc, int refEndLoc, int minScore){
jpayne@68 91 initialize(read.length, refEndLoc-refStartLoc+1);
jpayne@68 92
jpayne@68 93 //temporary, for finding a bug
jpayne@68 94 if(rows>maxRows || columns>maxColumns){
jpayne@68 95 throw new RuntimeException("rows="+rows+", maxRows="+maxRows+", cols="+columns+", maxCols="+maxColumns+"\n"+new String(read)+"\n");
jpayne@68 96 }
jpayne@68 97
jpayne@68 98 assert(rows<=maxRows) : "Check that values are in-bounds before calling this function: "+rows+", "+maxRows;
jpayne@68 99 assert(columns<=maxColumns) : "Check that values are in-bounds before calling this function: "+columns+", "+maxColumns;
jpayne@68 100
jpayne@68 101 assert(refStartLoc>=0) : "Check that values are in-bounds before calling this function: "+refStartLoc;
jpayne@68 102 assert(refEndLoc<ref.length) : "Check that values are in-bounds before calling this function: "+refEndLoc+", "+ref.length;
jpayne@68 103
jpayne@68 104 final int refOffset=refStartLoc-1;
jpayne@68 105 for(int row=1; row<=rows; row++){
jpayne@68 106
jpayne@68 107 final byte qBase=read[row-1];
jpayne@68 108 for(int col=1; col<=columns; col++){
jpayne@68 109
jpayne@68 110 final byte rBase=ref[refOffset+col];
jpayne@68 111
jpayne@68 112 final boolean match=(qBase==rBase);
jpayne@68 113 final boolean defined=(AminoAcid.isFullyDefinedAA(qBase) && AminoAcid.isFullyDefinedAA(rBase));
jpayne@68 114
jpayne@68 115 final int scoreFromDiag=packed[row-1][col-1];
jpayne@68 116 final int scoreFromDel=packed[row][col-1];
jpayne@68 117 final int scoreFromIns=packed[row-1][col];
jpayne@68 118
jpayne@68 119 final int diagScoreM=POINTS_MATCH;
jpayne@68 120 final int diagScoreS=POINTS_SUB;
jpayne@68 121 final int delScore=scoreFromDel+POINTS_DEL;
jpayne@68 122 final int insScore=scoreFromIns+POINTS_INS;
jpayne@68 123
jpayne@68 124 // final int diagScore=scoreFromDiag+(defined ? (match ? diagScoreM : diagScoreS) : POINTS_NOREF);
jpayne@68 125 int diagScore=(match ? diagScoreM : diagScoreS);
jpayne@68 126 diagScore=scoreFromDiag+(defined ? diagScore : POINTS_NOREF);
jpayne@68 127
jpayne@68 128 int score=diagScore>=delScore ? diagScore : delScore;
jpayne@68 129 score=score>=insScore ? score : insScore;
jpayne@68 130
jpayne@68 131 packed[row][col]=score;
jpayne@68 132 }
jpayne@68 133 //iterationsUnlimited+=columns;
jpayne@68 134 }
jpayne@68 135
jpayne@68 136
jpayne@68 137 int maxCol=-1;
jpayne@68 138 int maxState=-1;
jpayne@68 139 int maxStart=-1;
jpayne@68 140 int maxScore=Integer.MIN_VALUE;
jpayne@68 141
jpayne@68 142 for(int col=1; col<=columns; col++){
jpayne@68 143 int x=packed[rows][col];
jpayne@68 144 if(x>maxScore){
jpayne@68 145 maxScore=x;
jpayne@68 146 maxCol=col;
jpayne@68 147
jpayne@68 148 // assert(rows-1<read.length) : (rows-1)+", "+read.length;
jpayne@68 149 // assert(refOffset+col<ref.length) : refOffset+", "+col+", "+ref.length;
jpayne@68 150 maxState=getState(rows, col, read[rows-1], ref[refOffset+col]);
jpayne@68 151 maxStart=x;
jpayne@68 152 }
jpayne@68 153 }
jpayne@68 154
jpayne@68 155 // System.err.println("Returning "+rows+", "+maxCol+", "+maxState+", "+maxScore+"; minScore="+minScore);
jpayne@68 156 return maxScore<minScore ? null : new int[] {rows, maxCol, maxState, maxScore, maxStart};
jpayne@68 157 }
jpayne@68 158
jpayne@68 159 int getState(int row, int col, byte q, byte r){//zxvzxcv TODO: Fix - needs to find max
jpayne@68 160 final boolean match=(q==r);
jpayne@68 161 final boolean defined=(AminoAcid.isFullyDefinedAA(q) && AminoAcid.isFullyDefinedAA(r));
jpayne@68 162
jpayne@68 163 final int scoreFromDiag=packed[row-1][col-1];
jpayne@68 164 final int scoreFromDel=packed[row][col-1];
jpayne@68 165 final int scoreFromIns=packed[row-1][col];
jpayne@68 166 // final int score=packed[row][col];
jpayne@68 167
jpayne@68 168 final int diagScoreM=POINTS_MATCH;
jpayne@68 169 final int diagScoreS=POINTS_SUB;
jpayne@68 170 final int delScore=scoreFromDel+POINTS_DEL;
jpayne@68 171 final int insScore=scoreFromIns+POINTS_INS;
jpayne@68 172
jpayne@68 173 final int diagScore=scoreFromDiag+(defined ? (match ? diagScoreM : diagScoreS) : POINTS_NOREF);
jpayne@68 174
jpayne@68 175 // int score2=diagScore>=delScore ? diagScore : delScore;
jpayne@68 176 // score2=score>=insScore ? score : insScore;
jpayne@68 177
jpayne@68 178 // assert(score==score2) : score+", "+score2;
jpayne@68 179
jpayne@68 180 if(diagScore>=delScore && diagScore>=insScore){
jpayne@68 181 return defined ? match ? MODE_MATCH : MODE_SUB : MODE_N;
jpayne@68 182 }else if(delScore>=insScore){
jpayne@68 183 return MODE_DEL;
jpayne@68 184 }
jpayne@68 185 return MODE_INS;
jpayne@68 186 }
jpayne@68 187
jpayne@68 188 /** Generates the match string */
jpayne@68 189 @Override
jpayne@68 190 public final byte[] traceback(byte[] query, byte[] ref, int refStartLoc, int refEndLoc, int row, int col, int state){
jpayne@68 191 // assert(false);
jpayne@68 192 assert(refStartLoc<=refEndLoc) : refStartLoc+", "+refEndLoc;
jpayne@68 193 assert(row==rows);
jpayne@68 194
jpayne@68 195 byte[] out=new byte[row+col-1]; //TODO if an out of bound crash occurs, try removing the "-1".
jpayne@68 196 int outPos=0;
jpayne@68 197
jpayne@68 198 // assert(state==(packed[row][col]&MODEMASK));
jpayne@68 199
jpayne@68 200 while(row>0 && col>0){
jpayne@68 201 byte q=query[row-1];
jpayne@68 202 byte r=ref[refStartLoc+col-1];
jpayne@68 203 boolean defined=(AminoAcid.isFullyDefinedAA(q) && AminoAcid.isFullyDefinedAA(r));
jpayne@68 204 state=getState(row, col, q, r);
jpayne@68 205 // assert(defined) : state+", "+(int)q+", "+(int)r+", "+new String(query);
jpayne@68 206 // assert(state!=MODE_N) : state+", "+Character.toString(q)+", "+Character.toString(r)+", "+new String(query);
jpayne@68 207 if(state==MODE_MATCH){
jpayne@68 208 col--;
jpayne@68 209 row--;
jpayne@68 210 out[outPos]=defined ? (byte)'m' : (byte)'N';
jpayne@68 211 }else if(state==MODE_SUB){
jpayne@68 212 col--;
jpayne@68 213 row--;
jpayne@68 214 out[outPos]=defined ? (byte)'S' : (byte)'N';
jpayne@68 215 }else if(state==MODE_N){
jpayne@68 216 col--;
jpayne@68 217 row--;
jpayne@68 218 out[outPos]='N';
jpayne@68 219 }else if(state==MODE_DEL){
jpayne@68 220 col--;
jpayne@68 221 out[outPos]='D';
jpayne@68 222 }else if(state==MODE_INS){
jpayne@68 223 row--;
jpayne@68 224 // out[outPos]='I';
jpayne@68 225 if(col>=0 && col<columns){
jpayne@68 226 out[outPos]='I';
jpayne@68 227 }else{
jpayne@68 228 out[outPos]='C';
jpayne@68 229 col--;
jpayne@68 230 }
jpayne@68 231 }else{
jpayne@68 232 assert(false) : state;
jpayne@68 233 }
jpayne@68 234 outPos++;
jpayne@68 235 }
jpayne@68 236
jpayne@68 237 assert(row==0 || col==0);
jpayne@68 238 if(col!=row){//Not sure what this is doing
jpayne@68 239 while(row>0){
jpayne@68 240 out[outPos]='C';
jpayne@68 241 outPos++;
jpayne@68 242 row--;
jpayne@68 243 col--;
jpayne@68 244 }
jpayne@68 245 if(col>0){
jpayne@68 246 //do nothing
jpayne@68 247 }
jpayne@68 248 }
jpayne@68 249
jpayne@68 250 //Shrink and reverse the string
jpayne@68 251 byte[] out2=new byte[outPos];
jpayne@68 252 for(int i=0; i<outPos; i++){
jpayne@68 253 out2[i]=out[outPos-i-1];
jpayne@68 254 }
jpayne@68 255 out=null;
jpayne@68 256
jpayne@68 257 return out2;
jpayne@68 258 }
jpayne@68 259
jpayne@68 260 @Override
jpayne@68 261 /** Generates identity;
jpayne@68 262 * fills 'extra' with {match, sub, del, ins, N, clip} if present */
jpayne@68 263 public float tracebackIdentity(byte[] query, byte[] ref, int refStartLoc, int refEndLoc, int row, int col, int state, int[] extra){
jpayne@68 264
jpayne@68 265 // assert(false);
jpayne@68 266 assert(refStartLoc<=refEndLoc) : refStartLoc+", "+refEndLoc;
jpayne@68 267 assert(row==rows);
jpayne@68 268
jpayne@68 269 // assert(state==(packed[row][col]&MODEMASK));
jpayne@68 270 int match=0, sub=0, del=0, ins=0, noref=0, clip=0;
jpayne@68 271
jpayne@68 272 while(row>0 && col>0){
jpayne@68 273 byte q=query[row-1];
jpayne@68 274 byte r=ref[refStartLoc+col-1];
jpayne@68 275 boolean defined=(AminoAcid.isFullyDefinedAA(q) && AminoAcid.isFullyDefinedAA(r));
jpayne@68 276 state=getState(row, col, q, r);
jpayne@68 277 if(state==MODE_MATCH){
jpayne@68 278 col--;
jpayne@68 279 row--;
jpayne@68 280 match+=(defined ? 1 : 0);
jpayne@68 281 noref+=(defined ? 0 : 1);
jpayne@68 282 }else if(state==MODE_SUB){
jpayne@68 283 col--;
jpayne@68 284 row--;
jpayne@68 285 sub+=(defined ? 1 : 0);
jpayne@68 286 noref+=(defined ? 0 : 1);
jpayne@68 287 }else if(state==MODE_N){
jpayne@68 288 col--;
jpayne@68 289 row--;
jpayne@68 290 noref++;
jpayne@68 291 }else if(state==MODE_DEL){
jpayne@68 292 col--;
jpayne@68 293 del++;
jpayne@68 294 }else if(state==MODE_INS){
jpayne@68 295 row--;
jpayne@68 296 boolean edge=(col<=1 || col>=columns);
jpayne@68 297 ins+=(edge ? 0 : 1);
jpayne@68 298 clip+=(edge ? 1 : 0);
jpayne@68 299 }else{
jpayne@68 300 assert(false) : state;
jpayne@68 301 }
jpayne@68 302 }
jpayne@68 303
jpayne@68 304 assert(row==0 || col==0);
jpayne@68 305 if(col!=row){//Not sure what this is doing
jpayne@68 306 while(row>0){
jpayne@68 307 clip++;
jpayne@68 308 row--;
jpayne@68 309 col--;
jpayne@68 310 }
jpayne@68 311 if(col>0){
jpayne@68 312 //do nothing
jpayne@68 313 }
jpayne@68 314 }
jpayne@68 315
jpayne@68 316 if(extra!=null){
jpayne@68 317 assert(extra.length==5);
jpayne@68 318 extra[0]=match;
jpayne@68 319 extra[1]=sub;
jpayne@68 320 extra[2]=del;
jpayne@68 321 extra[3]=ins;
jpayne@68 322 extra[4]=noref;
jpayne@68 323 extra[5]=clip;
jpayne@68 324 }
jpayne@68 325
jpayne@68 326 float len=match+sub+ins+del+noref*0.1f;
jpayne@68 327 float id=match/Tools.max(1.0f, len);
jpayne@68 328 return id;
jpayne@68 329 }
jpayne@68 330
jpayne@68 331 /** Generates identity;
jpayne@68 332 * fills 'extra' with {match, sub, del, ins, N, clip} if present */
jpayne@68 333 public float tracebackIdentityAmino(byte[] query, byte[] ref, int refStartLoc, int refEndLoc, int row, int col, int state, int[] extra){
jpayne@68 334
jpayne@68 335 // assert(false);
jpayne@68 336 assert(refStartLoc<=refEndLoc) : refStartLoc+", "+refEndLoc;
jpayne@68 337 assert(row==rows);
jpayne@68 338
jpayne@68 339 // assert(state==(packed[row][col]&MODEMASK));
jpayne@68 340 int match=0, sub=0, del=0, ins=0, noref=0, clip=0;
jpayne@68 341
jpayne@68 342 while(row>0 && col>0){
jpayne@68 343 byte q=query[row-1];
jpayne@68 344 byte r=ref[refStartLoc+col-1];
jpayne@68 345 boolean defined=(AminoAcid.isFullyDefinedAA(q) && AminoAcid.isFullyDefinedAA(r));
jpayne@68 346 state=getState(row, col, q, r);
jpayne@68 347 if(state==MODE_MATCH){
jpayne@68 348 col--;
jpayne@68 349 row--;
jpayne@68 350 match+=(defined ? 1 : 0);
jpayne@68 351 noref+=(defined ? 0 : 1);
jpayne@68 352 }else if(state==MODE_SUB){
jpayne@68 353 col--;
jpayne@68 354 row--;
jpayne@68 355 sub+=(defined ? 1 : 0);
jpayne@68 356 noref+=(defined ? 0 : 1);
jpayne@68 357 }else if(state==MODE_N){
jpayne@68 358 col--;
jpayne@68 359 row--;
jpayne@68 360 noref++;
jpayne@68 361 }else if(state==MODE_DEL){
jpayne@68 362 col--;
jpayne@68 363 del++;
jpayne@68 364 }else if(state==MODE_INS){
jpayne@68 365 row--;
jpayne@68 366 boolean edge=(col<=1 || col>=columns);
jpayne@68 367 ins+=(edge ? 0 : 1);
jpayne@68 368 clip+=(edge ? 1 : 0);
jpayne@68 369 }else{
jpayne@68 370 assert(false) : state;
jpayne@68 371 }
jpayne@68 372 }
jpayne@68 373
jpayne@68 374 assert(row==0 || col==0);
jpayne@68 375 if(col!=row){//Not sure what this is doing
jpayne@68 376 while(row>0){
jpayne@68 377 clip++;
jpayne@68 378 row--;
jpayne@68 379 col--;
jpayne@68 380 }
jpayne@68 381 if(col>0){
jpayne@68 382 //do nothing
jpayne@68 383 }
jpayne@68 384 }
jpayne@68 385
jpayne@68 386 if(extra!=null){
jpayne@68 387 assert(extra.length==5);
jpayne@68 388 extra[0]=match;
jpayne@68 389 extra[1]=sub;
jpayne@68 390 extra[2]=del;
jpayne@68 391 extra[3]=ins;
jpayne@68 392 extra[4]=noref;
jpayne@68 393 extra[5]=clip;
jpayne@68 394 }
jpayne@68 395
jpayne@68 396 float len=match+sub+ins+del+noref*0.1f;
jpayne@68 397 float id=match/Tools.max(1.0f, len);
jpayne@68 398 return id;
jpayne@68 399 }
jpayne@68 400
jpayne@68 401 /** @return {score, bestRefStart, bestRefStop} */
jpayne@68 402 @Override
jpayne@68 403 public final int[] score(final byte[] read, final byte[] ref, final int refStartLoc, final int refEndLoc,
jpayne@68 404 final int maxRow, final int maxCol, final int maxState/*, final int maxScore, final int maxStart*/){
jpayne@68 405
jpayne@68 406 int row=maxRow;
jpayne@68 407 int col=maxCol;
jpayne@68 408 int state=maxState;
jpayne@68 409
jpayne@68 410 assert(maxState>=0 && maxState<packed.length) :
jpayne@68 411 maxState+", "+maxRow+", "+maxCol+"\n"+new String(read)+"\n"+toString(ref, refStartLoc, refEndLoc);
jpayne@68 412 assert(maxRow>=0 && maxRow<packed.length) :
jpayne@68 413 maxState+", "+maxRow+", "+maxCol+"\n"+new String(read)+"\n"+toString(ref, refStartLoc, refEndLoc);
jpayne@68 414 assert(maxCol>=0 && maxCol<packed[0].length) :
jpayne@68 415 maxState+", "+maxRow+", "+maxCol+"\n"+new String(read)+"\n"+toString(ref, refStartLoc, refEndLoc);
jpayne@68 416
jpayne@68 417 int score=packed[maxRow][maxCol]; //Or zero, if it is to be recalculated
jpayne@68 418
jpayne@68 419 if(row<rows){
jpayne@68 420 int difR=rows-row;
jpayne@68 421 int difC=columns-col;
jpayne@68 422
jpayne@68 423 while(difR>difC){
jpayne@68 424 score+=POINTS_NOREF;
jpayne@68 425 difR--;
jpayne@68 426 }
jpayne@68 427
jpayne@68 428 row+=difR;
jpayne@68 429 col+=difR;
jpayne@68 430
jpayne@68 431 }
jpayne@68 432
jpayne@68 433 assert(refStartLoc<=refEndLoc);
jpayne@68 434 assert(row==rows);
jpayne@68 435
jpayne@68 436
jpayne@68 437 final int bestRefStop=refStartLoc+col-1;
jpayne@68 438
jpayne@68 439 while(row>0 && col>0){
jpayne@68 440 final byte q=read[row-1];
jpayne@68 441 final byte r=ref[refStartLoc+col-1];
jpayne@68 442 // final boolean defined=(AminoAcid.isFullyDefinedAA(q) && AminoAcid.isFullyDefinedAA(r));
jpayne@68 443 state=getState(row, col, q, r);
jpayne@68 444 if(state==MODE_MATCH){
jpayne@68 445 col--;
jpayne@68 446 row--;
jpayne@68 447 }else if(state==MODE_SUB){
jpayne@68 448 col--;
jpayne@68 449 row--;
jpayne@68 450 }else if(state==MODE_N){
jpayne@68 451 col--;
jpayne@68 452 row--;
jpayne@68 453 }else if(state==MODE_DEL){
jpayne@68 454 col--;
jpayne@68 455 }else if(state==MODE_INS){
jpayne@68 456 row--;
jpayne@68 457 }else{
jpayne@68 458 assert(false) : state;
jpayne@68 459 }
jpayne@68 460 }
jpayne@68 461 // assert(false) : row+", "+col;
jpayne@68 462 if(row>col){
jpayne@68 463 col-=row;
jpayne@68 464 }
jpayne@68 465
jpayne@68 466 final int bestRefStart=refStartLoc+col;
jpayne@68 467
jpayne@68 468 // System.err.println("t2\t"+score+", "+maxScore+", "+maxStart+", "+bestRefStart);
jpayne@68 469 int[] rvec;
jpayne@68 470 if(bestRefStart<refStartLoc || bestRefStop>refEndLoc){ //Suggest extra padding in cases of overflow
jpayne@68 471 int padLeft=Tools.max(0, refStartLoc-bestRefStart);
jpayne@68 472 int padRight=Tools.max(0, bestRefStop-refEndLoc);
jpayne@68 473 rvec=new int[] {score, bestRefStart, bestRefStop, padLeft, padRight};
jpayne@68 474 }else{
jpayne@68 475 rvec=new int[] {score, bestRefStart, bestRefStop};
jpayne@68 476 }
jpayne@68 477 return rvec;
jpayne@68 478 }
jpayne@68 479
jpayne@68 480
jpayne@68 481 /** Will not fill areas that cannot match minScore.
jpayne@68 482 * @return {score, bestRefStart, bestRefStop} */
jpayne@68 483 @Override
jpayne@68 484 public final int[] fillAndScoreLimited(byte[] read, byte[] ref, int refStartLoc, int refEndLoc, int minScore){
jpayne@68 485 int a=Tools.max(0, refStartLoc);
jpayne@68 486 int b=Tools.min(ref.length-1, refEndLoc);
jpayne@68 487 assert(b>=a);
jpayne@68 488
jpayne@68 489 if(b-a>=maxColumns){
jpayne@68 490 System.err.println("Warning: Max alignment columns exceeded; restricting range. "+(b-a+1)+" > "+maxColumns);
jpayne@68 491 assert(false) : refStartLoc+", "+refEndLoc;
jpayne@68 492 b=Tools.min(ref.length-1, a+maxColumns-1);
jpayne@68 493 }
jpayne@68 494 int[] max=fillLimited(read, ref, a, b, minScore);
jpayne@68 495 // return max==null ? null : new int[] {max[3], 0, max[1]};
jpayne@68 496
jpayne@68 497 int[] score=(max==null ? null : score(read, ref, a, b, max[0], max[1], max[2]/*, max[3], max[4]*/));
jpayne@68 498
jpayne@68 499 return score;
jpayne@68 500 }
jpayne@68 501
jpayne@68 502 public static final String toString(byte[] ref, int startLoc, int stopLoc){
jpayne@68 503 StringBuilder sb=new StringBuilder(stopLoc-startLoc+1);
jpayne@68 504 for(int i=startLoc; i<=stopLoc; i++){sb.append((char)ref[i]);}
jpayne@68 505 return sb.toString();
jpayne@68 506 }
jpayne@68 507
jpayne@68 508 // public static int calcDelScore(int len){
jpayne@68 509 // if(len<=0){return 0;}
jpayne@68 510 // int score=POINTS_DEL;
jpayne@68 511 // if(len>1){
jpayne@68 512 // score+=(len-1)*POINTS_DEL2;
jpayne@68 513 // }
jpayne@68 514 // return score;
jpayne@68 515 // }
jpayne@68 516
jpayne@68 517 // public int maxScoreByIdentity(int len, float identity){
jpayne@68 518 // assert(identity>=0 && identity<=1);
jpayne@68 519 // return (int)(len*(identity*POINTS_MATCH+(1-identity)*POINTS_SUB));
jpayne@68 520 // }
jpayne@68 521
jpayne@68 522 @Override
jpayne@68 523 public int minScoreByIdentity(int len, float identity){
jpayne@68 524 assert(identity>=0 && identity<=1);
jpayne@68 525
jpayne@68 526 int a=(int)(len*(identity*POINTS_MATCH+(1-identity)*POINTS_SUB));
jpayne@68 527 int b=(int)(len*(identity*POINTS_MATCH+(1-identity)*POINTS_INS));
jpayne@68 528 int c=(int)(len*(1*POINTS_MATCH+((1/(Tools.max(identity, 0.000001f)))-1)*POINTS_DEL));
jpayne@68 529 return Tools.min(a, b, c);
jpayne@68 530 }
jpayne@68 531
jpayne@68 532 private static int calcDelScore(int len){
jpayne@68 533 if(len<=0){return 0;}
jpayne@68 534 int score=POINTS_DEL*len;
jpayne@68 535 return score;
jpayne@68 536 }
jpayne@68 537
jpayne@68 538 @Override
jpayne@68 539 public int rows(){return rows;}
jpayne@68 540 @Override
jpayne@68 541 public int columns(){return columns;}
jpayne@68 542
jpayne@68 543
jpayne@68 544 private int maxRows;
jpayne@68 545 private int maxColumns;
jpayne@68 546
jpayne@68 547 private int[][] packed;
jpayne@68 548
jpayne@68 549 public static final int MAX_SCORE=Integer.MAX_VALUE-2000;
jpayne@68 550 public static final int MIN_SCORE=0-MAX_SCORE; //Keeps it 1 point above "BAD".
jpayne@68 551
jpayne@68 552 //For some reason changing MODE_DEL from 1 to 0 breaks everything
jpayne@68 553 private static final byte MODE_DEL=1;
jpayne@68 554 private static final byte MODE_INS=2;
jpayne@68 555 private static final byte MODE_SUB=3;
jpayne@68 556 private static final byte MODE_MATCH=4;
jpayne@68 557 private static final byte MODE_N=5;
jpayne@68 558
jpayne@68 559 public static final int POINTS_NOREF=-15;
jpayne@68 560 public static final int POINTS_MATCH=100;
jpayne@68 561 public static final int POINTS_SUB=-50;
jpayne@68 562 public static final int POINTS_INS=-121;
jpayne@68 563 public static final int POINTS_DEL=-111;
jpayne@68 564
jpayne@68 565 // public static final int POINTS_NOREF=-100000;
jpayne@68 566 // public static final int POINTS_MATCH=100;
jpayne@68 567 // public static final int POINTS_SUB=-100;
jpayne@68 568 // public static final int POINTS_INS=-100;
jpayne@68 569 // public static final int POINTS_DEL=-100;
jpayne@68 570
jpayne@68 571 public static final int BAD=MIN_SCORE-1;
jpayne@68 572
jpayne@68 573 private int rows;
jpayne@68 574 private int columns;
jpayne@68 575
jpayne@68 576 // public long iterationsLimited=0;
jpayne@68 577 // public long iterationsUnlimited=0;
jpayne@68 578
jpayne@68 579 public boolean verbose=false;
jpayne@68 580 public boolean verbose2=false;
jpayne@68 581
jpayne@68 582 }