devbox
繁依Fanyi0
这个屌丝很懒，什么也没留下！
关注作者
热门标签
热门文章
当前位置:   article > 正文

一份开源的果蝇算法C++源代码_果蝇算法c代码

作者：繁依Fanyi0 | 2024-04-02 16:56:24

赞

踩

果蝇算法c代码

仅供参考

编译此代码，您需要系统上已经安装 了“Gnu Scientific Library” （GSL）。

  1.  
  2.  
  3. #include <string>
  4. #include <iostream>
  5. #include <fstream>
  6. #include <vector>
  7. #include <sstream>
  8. #include <math.h>
  9. #include <cstdlib>
  10. #include <map>
  11. #include <gsl/gsl_randist.h>
  12. #include <gsl/gsl_cdf.h>
  13. //#include <gd.h>
  14.  
  15. //#define DEBUG1
  16. //#define DEBUG2
  17. //#define DEBUG3
  18. //#define DEBUG4
  19. //#define DEBUG5
  20. //#define DEBUG6
  21. //#define DEBUG7
  22. //#define DEBUG8
  23. //#define DEBUG9
  24. //#define MATRIX_TRANSLATOR
  25. //#define BOUND_OLIGOS
  26.  
  27. using namespace std;
  28.  
  29. //CONSTANTS DEFINITIONS
  30. typedef double m_point;
  31. const int ALPHABET_SIZE = 4, SWITCH_TO_Z_VALUE = 500, MAX_DOTS = 150000, DA_TOP = 20;
  32. const char ALPHABET[4] = {'A','C','G','T'};
  33. const char C_ALPHABET[4] = {'T','G','C','A'};
  34. const m_point MIN_ADD_FREQ = 0.01, INF = 1000000000, CONSERVATION_THRESHOLD = 0.75;
  35. const char *matrix_files = "*.pro *.pfm";
  36. enum {A,C,G,T};
  37. char VI['T' + 1], rc_VI['T' + 1];
  38. //*fasta_files = "*.fasta", 
  39. const char *m_start_identifier = "P0", *m_start_identifier_for_free_transfac = "PO", *m_data_start_identifier = "AC";
  40. enum {NO_FILES, BAD_MATRIX_FILE, MATRIX_NOT_ALLOCATED, MISSING_FASTA_FILE, BAD_FASTA_SEQUENCE, 
  41.       NO_MATRIX_FILE, BAD_MATRIX, NO_JASPAR_MATRIX_LIST, JASPAR_FILE_NOT_IN_LIST, CANT_COMPUTE_SCORES, READING_OUT_OF_MATRIX, 
  42.       NEED_COMPLETE_MATRIX, NO_BACKGROUND_AVAILABLE, NO_SEQUENCE, NO_TFBS, CANT_DO_POSITIONAL, SEQUENCE_SHORTER_THAN_WINDOW, BAD_BIG_VALUES, CANTOPENOUTPUT, TOO_MANY_EDGES, NO_BOF, BAD_FILE_FORMAT, NO_INDEX_FOUND, EVEN_PROBE, MISSING_CHIP_FILE, INCONSISTENT_CHIP, NO_MATRIX_FOR_DISTANCE_ANALYSIS}; //ERROR CODES
  43. const m_point step = 0.025; //bins step!!
  44. //END
  45.  
  46. //OBJECTS DECLARATION
  47. #ifdef  BOUND_OLIGOS
  48. class bound_oligos
  49. {
  50. 	private:
  51. 	void build_matrix();
  52. 	void build_pos_cor_matrix();
  53. 	void w2_vector(vector<string>*);
  54. 	int char_to_index(char);
  55. 	int observed_cooccurrences(int, char, int, char);
  56. 	vector<string> raw_mat;
  57. 	m_point **pos_cor_mat;
  58. 	string NAME;
  59. 	int **mat, x_size, y_size;
  60. 	public:
  61. 	vector<string> assign(vector<string>);
  62. 	string show_pos_cor_mat();
  63. };
  64. #endif
  65.  
  66. class tfbs
  67. {
  68. 	private:
  69. 	bool raw_tfbsdata_parser();
  70. 	void j_raw_tfbsdata_parser();
  71. 	bool raw_matrix_parser();
  72. 	bool j_raw_matrix_parser();
  73. 	bool matrix_normalizer();
  74. 	void matrix_to_log();
  75. 	void matrix_min_max();
  76. 	void bns_sum();
  77. 	void matrix_average();
  78. 	void build_conservation_vector();
  79. 	vector<bool> v_conservation;
  80. 	vector<string> raw_matrix;
  81. 	vector<int> bns;
  82. 	vector<m_point> row;
  83. 	vector<string> raw_data;
  84. 	m_point bin_freq(int);
  85. 	string j_raw_data;
  86. 	m_point **matrix;
  87. 	m_point *mat_avg;
  88. 	m_point avg_row;
  89. 	m_point stddev_row;
  90. 	string file_name;
  91. //	string raw_data;
  92. 	string ID;
  93. 	string NAME;
  94. 	string AC;
  95. 	int rowsize;
  96. 	int x_size;
  97. 	int bin_sum;
  98. 	m_point max_score;
  99. 	m_point min_score;
  100. 	bool matrix_alloc;
  101. 	bool norm_sec_step;
  102. 	public:
  103. 	bool assign(const char*);
  104. 	bool j_assign(const char*);
  105. 	bool w2_assign(vector<string>);
  106. 	#ifdef BOUND_OLIGOS
  107. 	bound_oligos bo;
  108. 	#endif
  109. 	string name();
  110. 	string id();
  111. 	string file();
  112. 	string matrix_id();
  113. 	string ac();
  114. 	string show_raw_matrix();
  115. 	#ifdef MATRIX_TRANSLATOR
  116. 	string matrix_translator();
  117. 	#endif
  118. 	string show_matrix();
  119. 	int get_bin_sum();
  120. 	m_point max();
  121. 	m_point min();
  122. 	void gen_bns();
  123. 	string show_bns();
  124. 	string show_bns_line();
  125. 	string show_conservation_vector(int);
  126. 	int m_length();
  127. 	bool assign_short_matrix(string);
  128. 	void row_average();
  129. 	void row_stddev();
  130. 	m_point get_value(int,int);
  131. 	m_point get_matrix_average(int);
  132. 	void row_push(m_point);
  133. 	void row_v_push(vector<m_point>);
  134. 	int row_size();
  135. 	int row_actual_size();
  136. 	m_point row_get(int);
  137. 	vector<m_point> row_get();
  138. 	m_point row_avg_get();
  139. 	m_point row_stddev_get();
  140. 	m_point bin_prob(int);
  141. 	vector<m_point>::iterator row_iter();
  142. 	unsigned int bin_hm(int);
  143. 	vector<int> bin_vector();
  144. //	void draw_background();
  145. };
  146. class sequence
  147. {
  148. 	private:
  149. 	vector<vector<m_point> > seq_rel_scores;
  150. 	vector<vector<char > > seq_r_strand;
  151. 	vector<m_point> max_score;
  152. 	vector<m_point> rel_score;
  153. 	vector<m_point> rel_score2;
  154. 	vector<int> max_pos;
  155. 	vector<int> max_pos2;
  156. 	vector<char> max_strand;
  157. 	string NAME;
  158. 	string SEQ;
  159. 	string WRAP_NAME;
  160. 	string ID;
  161. 	string CHR;
  162. 	unsigned long int ABS_START;
  163. 	unsigned long int ABS_END;
  164. 	char STRAND;
  165. 	virtual void SCAN(vector<tfbs>::iterator,bool);
  166. 	void SCAN_SS(vector<tfbs>::iterator,bool);
  167. 	void store_complete_seq_scores(vector<m_point>::iterator,vector<m_point>::iterator,int,vector<tfbs>::iterator);
  168. 	void store_complete_seq_scores_ss(vector<m_point>::iterator,int,vector<tfbs>::iterator);
  169. //	void set_seq_rel_max_pos();
  170. 	bool absolute_position_parser();
  171. 	protected:
  172. 	bool seq_cleaner();
  173. 	int char_to_index(char);
  174. 	int comp_char_to_index(char);
  175. 	m_point max_v(vector<m_point>*,int*);
  176. 	m_point max_v2(vector<m_point>*,int*); 
  177. 	m_point sum_v(vector<m_point>*);
  178. 	public:
  179. 	m_point get_seq_rel_score_at_pos(int,int);
  180. 	m_point get_seq_rel_max_pos(int,int*,char*);
  181. 	unsigned long int get_abs_start();
  182. 	unsigned long int get_abs_end();
  183. 	string get_chr();
  184. 	char get_strand();
  185. 	bool assign(string);
  186. 	virtual void scan(vector<tfbs>*,int,bool);
  187. 	string name();
  188. 	string seq();
  189. 	string w_name();
  190. 	string get_id();
  191. 	m_point max(int);
  192. 	m_point rel(int);
  193. 	m_point rel2(int);
  194. 	int mpos(int);
  195. 	int mpos2(int);
  196. 	char mstrand(int);
  197. 	void reverse();
  198. };
  199.  
  200. class big_sequence : public sequence
  201. {
  202. 	private:
  203. 	vector<vector <m_point> > MAX_REL_SCORE;
  204. 	vector<vector <int> > MAX_POS;
  205. 	vector<vector <string> > MAX_MOTIF;
  206. 	vector<vector<char> > MOTIF_STRAND;
  207. 	void SCAN(vector<tfbs>::iterator,bool);
  208. 	public:
  209. //	vector<m_point> row_return(int);
  210. 	void scan(vector<tfbs>*,int,bool);
  211. 	string track_output(int);
  212. 	string inline_output();
  213. };
  214.  
  215. class query
  216. {
  217. 	private:
  218. //	m_point **correlation_matrix;
  219. 	int dam, bigst_seq_size;
  220. 	vector<sequence> QUERY;
  221. 	vector<tfbs> QTFBS;
  222. 	vector<m_point> ts;
  223. 	vector<m_point> zt;
  224. 	vector<m_point> bt;
  225. 	vector<m_point> min_binom_bins;
  226. 	vector<vector <m_point> > binom_bins;
  227. 	vector<int> seq_bin_num;
  228. 	vector<int> soil_pos;
  229. 	void distance_analysis();
  230. 	char comp_char(char);
  231. 	string revcomp(string, char);
  232. 	m_point t_den_calc(int,m_point,int,m_point);
  233. 	int max_v_pos(vector<m_point>);
  234. 	int min_v_pos(vector<m_point>);
  235. //	void build_correlation_matrix(vector<tfbs>*);
  236. //	void correlation_matrix_output(vector<tfbs>*,string);
  237. //	void correlation_output(vector<tfbs>*,string);
  238. //	void correlation_matrix_copy(m_point**, int);
  239. 	vector<int> matrix_line_max_pos(m_point*,int,int);
  240. 	vector<int> matrix_line_min_pos(m_point*,int,int);
  241. 	string positional_output();
  242. 	m_point seq_stats(int, m_point*);
  243. 	char flip_strand(char);
  244. //	void draw_distr_graph(vector<m_point>*, vector<int>*, int, string);
  245. 	public:
  246. 	void assign(vector<sequence>,vector<tfbs>);
  247. 	void scan();
  248. 	void t_stud_calc(vector<tfbs>*);
  249. 	void z_test(vector<tfbs>*);
  250. 	void b_test(vector<tfbs>*);
  251. 	void bins_stuff(vector<tfbs>*);
  252. //	void correlation(vector<tfbs>*);
  253. 	void pearson_corr();
  254. 	string binom_bins_line(int);
  255. 	void output(int,vector<tfbs>*);
  256. //	void img_output(vector<tfbs>*);
  257. 	void zvectors_output(vector<tfbs>*);
  258. };
  259.  
  260. class chip
  261. {
  262. 	private:
  263. 	vector<m_point> Log_Ratio;
  264. 	unsigned short int noe;
  265. 	vector<string> Chr;
  266. 	vector<m_point> exp_bg_means;
  267. 	vector<m_point> exp_fg_means;
  268. //	vector<m_point> bin_label;
  269. 	vector<unsigned int> bins;
  270. 	vector<unsigned int> Start;
  271. 	vector<unsigned int> Stop;
  272. 	vector<m_point> Score;
  273. 	string current_chr, current_chr_seq;
  274. 	ofstream fout;
  275. 	bool consistency_check(vector<string> *);
  276. 	void stack_processor(vector<vector<string> > *);
  277. 	void set_bins();
  278. 	void display_bins();
  279. 	void Exp_Push_Mean(vector<string>*);
  280. 	void get_current_chr_seq(string, string*);
  281. 	public:
  282. 	chip();
  283. };
  284.  
  285.  
  286. //END
  287.  
  288. //GLOBAL VARIABLES DECLARATION
  289. bool VERBOSE = false, BINS = false, COMPLETE_MATRIX = false, SHORT = false, SHOW_BINS_BINOM = false, USE_N = false, CORRELATION = false,
  290.      COMPLETE_CORRELATION_MATRIX = false, GENERATE_MATRIX = false, POSITIONAL =  false, USE_SUM = false, BIG_SEQ = false, PEARSON = false,
  291.      NO_BACKGROUND = false, DOUBLE_STRAND = true, IMG_OUTPUT = false, TRASH_N = false, ZVECTOR_OUTPUT = false, USE_T = false, 
  292.      SHOW_CONSERVATION_VECTOR = false, REVERSE_STRAND = false, BED_OUTPUT = false, DRAW_BACKGROUND = false, BUILD_INDEX = false, NO_HEADER = false, CHIP = false;
  293. int BIG_WINDOW = 500, BIG_STEP = 250, SEQ_EDGE_START = 0, SEQ_EDGE_END = 0, MIN_SEQ_LENGTH = 0, TSS_POS = 0;
  294. string q_idlist, idfile, queryfile, promfile, matrixfile, usethismatrix, usethismatrixname, matrixlist, trackbg, w2file, fasta_matrix_file, bound_oligos_file, bg_for_img, magic, index_file, distance_analysis_matrix;
  295. vector<string> chip_files;
  296. vector<m_point> bins;
  297. double TRACK_CUTOFF = 0, CHIP_CUTOFF = 0.05, DISTANCE_CUTOFF = 0.000001;
  298. unsigned int SPLIT[2] = {0,0}, PROBE = 3;
  299. //END
  300.  
  301. //FUNCTIONS DECLARATION 
  302. vector<string> get_file_names(const char *);
  303. void error_handler(int, string);
  304. vector<tfbs> v_tfbs_builder(vector<string>);
  305. vector<sequence> v_sequence_builder(vector<string>, bool);
  306. vector<big_sequence> v_big_sequence_builder(vector<string>, bool);
  307. vector<string> fasta_reader(const char*);
  308. vector<vector <string> > w2_mat_reader(string);
  309. vector<vector <string> > wil_mat_reader(string);
  310. void v_tfbs_add_bof(vector<tfbs>*);
  311. void generate_output_matrix(vector<tfbs>*, vector<sequence>*);
  312. void generate_output_bins(vector<tfbs>*, vector<sequence>*);
  313. void generate_output_big_sequence(vector<tfbs>*, vector<big_sequence>*);
  314. void generate_tfbs_rows(vector<tfbs>*, vector<sequence>*, bool);
  315. //void generate_tfbs_rows(vector<tfbs>*, vector<big_sequence>*);
  316. void generate_tfbs_rows_from_matrix(vector<tfbs>*);
  317. void generate_tfbs_rows_from_short_matrix(vector<tfbs>*,ifstream*);
  318. void bins_gen();
  319. void command_line_parser(int, char**);
  320. void min_seq_length_fixer(vector<tfbs>*);
  321. void idfile_handler(vector<sequence> *);
  322. string build_sequence_index(vector<sequence>*);
  323. string show_conservation(vector<tfbs>*);
  324. bool id_check(vector<string> *, vector<sequence>::iterator);
  325. bool qlist_check(string, vector<string>*);
  326. void display_help();
  327. //END
  328.  
  329. int main(int argc, char **argv)
  330. {
  331. 	vector<tfbs> TFBS;
  332. 	vector<sequence> SEQUENCE;
  333. 	vector<big_sequence> BIG_SEQUENCE;
  334. 	query Q;
  335.  
  336. 	VI['A'] = 0;
  337.         VI['C'] = 1;
  338.         VI['G'] = 2;
  339.         VI['T'] = 3;
  340.  
  341.         rc_VI['T'] = 0;
  342.         rc_VI['G'] = 1;
  343.         rc_VI['C'] = 2;
  344.         rc_VI['A'] = 3;
  345.  
  346. 	command_line_parser(argc, argv);
  347. 	bins_gen();
  348.  
  349. 	if(CHIP)
  350. 	{
  351. 		chip C;
  352. 	}
  353.  
  354. 	TFBS = v_tfbs_builder(get_file_names(matrix_files));
  355. 	min_seq_length_fixer(&TFBS);
  356.  
  357. 	if((int)promfile.size() == 0 && (int)matrixfile.size() == 0 && usethismatrix.size() == 0 && !PEARSON && (int)usethismatrixname.size() == 0 && distance_analysis_matrix.empty())
  358. 	//	SEQUENCE = v_sequence_builder(get_file_names(fasta_files),false);
  359. 		error_handler(NO_BACKGROUND_AVAILABLE,"");
  360.  
  361. 	else if((int)promfile.size() != 0 && (int)matrixfile.size() == 0)
  362. 	{
  363. 		vector<string> pfile;
  364. 		pfile.push_back(promfile);
  365.  
  366. 		if(!BIG_SEQ)
  367. 		{
  368. 			SEQUENCE = v_sequence_builder(pfile,false);
  369. 			
  370. 			if(!idfile.empty())
  371. 				idfile_handler(&SEQUENCE);
  372. 		}
  373. 		
  374. 		else
  375. 		{
  376. 			if(BIG_STEP > BIG_WINDOW || BIG_STEP < 1)
  377. 				error_handler(BAD_BIG_VALUES,"");
  378.  
  379. 			BIG_SEQUENCE = v_big_sequence_builder(pfile,false);	
  380. 		}
  381. 	}
  382.  
  383. 	if((int)queryfile.size() != 0 && BIG_SEQUENCE.size() == 0)
  384. 	{
  385. 		vector<string> qfile;
  386. 		vector<sequence> QUERY;
  387. 		qfile.push_back(queryfile);
  388. 		QUERY = v_sequence_builder(qfile,true);
  389. 		Q.assign(QUERY,TFBS);
  390. 		QUERY.clear();
  391. 		qfile.clear();
  392. 	}
  393.  
  394. 	else if(BIG_SEQUENCE.size() != 0)
  395. 	{
  396. 		cerr << "Processing..." << endl;
  397. 		for(int S = 0; S < BIG_SEQUENCE.size(); S++)
  398.                                 BIG_SEQUENCE[S].scan(&TFBS,S,false);
  399.  
  400. 		generate_output_big_sequence(&TFBS, &BIG_SEQUENCE);
  401. 	}
  402.  
  403.  
  404. 	if((int)matrixfile.size() == 0 && promfile.size() != 0)
  405. 	{
  406. 		for(int S = 0; S < SEQUENCE.size(); S++)
  407. 			SEQUENCE[S].scan(&TFBS,S,false);
  408.  
  409. 		generate_tfbs_rows(&TFBS, &SEQUENCE, false);
  410.  
  411. 	}
  412.  
  413. 	else if(matrixfile.size() != 0 && promfile.size() == 0)
  414. 		generate_tfbs_rows_from_matrix(&TFBS);
  415.  
  416. 	else if(matrixfile.size() == 0 && promfile.size() == 0)
  417. 		NO_BACKGROUND = true;
  418.  
  419. 	if((int)queryfile.size() != 0)
  420. 	{
  421. 		Q.scan();
  422.  
  423. 		if(PEARSON)
  424.   	              Q.pearson_corr();
  425.  
  426. 		if(!POSITIONAL && !NO_BACKGROUND)
  427. 		{
  428. 			Q.t_stud_calc(&TFBS);
  429. 			Q.bins_stuff(&TFBS);
  430. 			Q.z_test(&TFBS);
  431. 			Q.b_test(&TFBS);
  432. 		}
  433.  
  434. 		Q.output(TFBS[0].row_size(),&TFBS);
  435.  
  436. //		if(IMG_OUTPUT && !POSITIONAL)
  437. //			Q.img_output(&TFBS);
  438.  
  439. 		if(ZVECTOR_OUTPUT)
  440. 			Q.zvectors_output(&TFBS);
  441. 	}
  442.  
  443. //	if(CORRELATION && TFBS.size() > 1)
  444. //		Q.correlation(&TFBS);
  445.  
  446. 	if((int)matrixfile.size() == 0 && (GENERATE_MATRIX || queryfile.size() == 0))
  447. 		generate_output_matrix(&TFBS, &SEQUENCE);
  448.  
  449. 	else if((int)matrixfile.size() != 0 && !SHORT && (GENERATE_MATRIX || queryfile.size() == 0))
  450. 	{
  451. 		COMPLETE_MATRIX = false;
  452. 		promfile = matrixfile;
  453. 		generate_output_matrix(&TFBS, &SEQUENCE);
  454. 	}
  455.  
  456. 	if(BINS)
  457. 		generate_output_bins(&TFBS, &SEQUENCE);
  458.  
  459. 	cerr << endl;
  460.  
  461. 	exit(EXIT_SUCCESS);
  462. }
  463.  
  464. void display_help()
  465. {
  466. 	ifstream in("HELP.txt");
  467.  
  468. 	if(!in)
  469. 	{
  470. 		cerr << "\nCan't find file: \"HELP.txt\" in this folder. Please check your installation." << endl;
  471. 		exit(1);
  472. 	}
  473.  
  474. 	string line;
  475.  
  476. 	while(getline(in,line))
  477. 		cerr << line << endl;
  478.  
  479. 	in.close();
  480.  
  481. /*	cerr << endl << "SYNOPSIS" << endl
  482. 	     << "\tpscan -q multifastafile -p multifastafile [options]" << endl
  483. 	     << "\tpscan -p multifastafile [options]" << endl
  484. 	     << "\tpscan -q multifastafile -M matrixfile [options]" << endl << endl
  485. 	     << "OPTIONS" << endl
  486. 	     << "\t[-q file] | Specify the multifasta file containing the foreground sequences." << endl << endl
  487. 	     << "\t[-p file] | Specify the multifasta file containing the background sequences." << endl << endl 
  488. 	     << "\t[-m file] | Use it if the background data are already available in a file (see -g option)." << endl << endl 
  489. 	     << "\t[-M file] | Scan the foreground sequences using only the Jaspar/Transfac matrix file contained in the specified file." << endl << endl
  490. 	     << "\t[-l file] | Use the matrices contained in the file (for matrix file format see below)." << endl << endl 		
  491. 	     << "\t[-N name] | Use only the matrix with that name (usable only in association with -l)." << endl << endl
  492. 	     << "\t[-ss] | Perform single strand only analysis." << endl << endl
  493. 	     << "\t[-rs] | Perform single strand only analysis on the reverse strand." << endl << endl
  494. 	     << "\t[-split num1 num2] | Sequences are scanned only from position num1 and for num2 nucleotides." << endl << endl
  495. 	     << "\t[-trashn] | Discards sequences containing \"N\"." << endl << endl
  496. 	     << "\t[-n] | Oligos containing \"N\" will not be discarded. Instead a \"N\" will obtain an \"average\" score." << endl << endl
  497. 	     << "\t[-g] | If a background sequences file is used than a file will be written containing the data calculated" << endl << 
  498. 		"\t\tfor that background and the current set of matrices." << endl << 
  499.                 "\t\tFrom now on one can use that file (-m option) instead of the sequences file for faster calculations." << endl << endl
  500. 	     << "\t[-ui file] | An index of the background file will be used to avoid duplicated sequences." << endl<< endl	
  501. 	     << "\t[-bi] | Build an index of the background sequences file (to be used later with the -ui option)." << endl << 
  502.                 "\t\tThis is useful when you have duplicated sequences in your background that may introduce a bias in your results." << endl << endl
  503. 	     << "\t[-h] | Display this help." << endl << endl
  505. 	<< "NOTES" << endl
  506. 	<< "\tThe sequences to be used with Pscan have to be promoter sequences." << endl
  507.         << "\tTo obtain meaningful results it's critical that the background and the foreground sequences are consistent between them either in size" << endl
  508.         << "\tand in position (with respect to the transcription start site). For optimal results the foreground set should be a subset of the background set." << endl << endl
  510.         << "\tIf the \"-l\" option is not used Pscan will try to find Jaspar/Transfac matrix files in the current folder." << endl
  511.         << "\tJaspar files have \".pfm\" extension while Transfac ones have \".pro\" extension." << endl
  512.         << "\tIf Jaspar matrix files are used than a file called \"matrix_list.txt\" must be present in the same folder." << endl
  513.         << "\tThat file contains required info about the matrices in the \".pfm\" files." << endl << endl
  515.         << "\tFor info on how Pscan works pleare refer to the paper." << endl << endl
  517. 	<< "EXAMPLES" << endl << endl
  518. 	<< "1)\tpscan -p human_450_50.fasta -bi" << endl << endl
  520.         << "\tThis command will scan the file \"human_450_50.fasta\" using the matrices in the current folder." << endl
  521.         << "\tIt is handy to use that command the first time one uses a set of matrices with a given background sequences file." << endl
  522.         << "\tA file called human_450_50.short_matrix will be written and it can be used from now on every time you want to use" << endl
  523.         << "\tthe same background sequences with the same set of matrices.  A file called human_450_50.index will be written too" << endl
  524.         << "\tand it will be useful every time you will use the same background file." << endl << endl
  526. 	<< "2)\tpscan -q human_nfy_targets.fasta -m human_450_50.short_matrix -ui human_450_50.index" << endl << endl
  528.         << "\tThis command will scan the file human_nfy_targets.fasta searching for over-represented binding sites (with respect" << endl
  529.         << "\tto the preprocessed background contained in the \"human_450_50.short_matrix\" file) using the matrices in the current folder." << endl
  530. 	<< "\tPlease note that the query file \"human_nfy_targets.fasta\" must be a subset of the sequences contained in the " << endl
  531. 	<< "\tbackground file \"human_450_50.fasta\"" 
  532.         << "\tin order to use the index file with the \"-ui\" option." << endl << "\tThis means that both the sequences and their FASTA headers used" << endl
  533. 	<< "\tin the query file must appear" 
  534.         << "\tin the background file as well." << endl << "\tUsing the \"-ui\" option when the sequences contained in the query file are not a subset of" << endl
  535. 	<< "\tthe background file will" 
  536.         << "have undefined/unpredictable outcomes."
  537.         << "\tThe output will be a file called \"human_nfy_targets.fasta.res\" where you will find all the used matrices sorted by ascending P-value." << endl
  538.         << "\tThe lower the P-value obtained by a matrix, the higher are the chances that the transcription factor associated to that matrix" << endl
  539.         << "\tis a regulator of the input promoter sequences." << endl
  540.         << "\tThe fields of the output are the following: \"Transcription Factor Name\", \"Matrix ID\", \"Z Score\", \"Pvalue\", \"Foreground Average\", \"Background Average\"." << endl << endl
  542. 	<< "3)\tpscan -q human_nfy_targets.fasta -M MA0108.pfm" << endl << endl
  544.         << "\tThis command will scan the sequences file \"human_nfy_targets.fasta\" using the matrix contained in \"MA0108.pfm\"." << endl
  545.         << "\tThe result will be written in a file called \"human_nfy_targets.fasta.ris\" where you will find the sequences in input" << endl
  546.         << "\tsorted by a descending score (between 1 and 0). The higher the score, the better is the oligo found with respect to the used matrix." << endl
  547.         << "\tThe fields of the output are the following: \"Sequence Header\", \"Score\", \"Position from the end of sequence\", \"Oligo that obtained the score\"," << endl
  548.         << "\tStrand where the oligo was found\"." << endl << endl
  550. 	<< "4)\tpscan -p human_450_50.fasta -bi -l matrixfile.wil" << endl << endl
  552.         << "\tThis command is like Example #1 with the difference that the matrices set to be used is the one contained in the \"matrixfile.wil\" file." << endl
  553.         << "\tPlease look at the \"example_matrix_file.wil\" file included in this Pscan distribution to see the correct format for matrices file." << endl << endl
  555. 	<< "5)\tpscan -q human_nfy_targets.fasta -l matrixfile.wil -N MATRIX1" << endl << endl
  557.         << "\tThis command is like Example #3 but it will use the matrix called \"MATRIX1\" contained in the \"matrixfile.wil\" file." << endl << endl;
  558. */
  559. 	exit(1);
  560. }
  561.  
  562. void command_line_parser(int argc, char **argv)
  563. {
  564.         if(argc == 1)
  565. 		display_help();
  566.  
  567.         for(int i = 1; i < argc; i++)
  568.         {
  569.                 string buf = argv[i];
  570.  
  571.                 if(buf == "-q")
  572.                 {
  573.                         if(i < argc - 1)
  574.                                 queryfile = argv[++i];
  575.                         continue;
  576.                 }
  577.  
  578. 		if(buf == "-h")
  579. 		{
  580. 			display_help();
  581. 			continue;	
  582. 		}
  583. /*		if(buf == "-Q")
  584. 		{
  585. 			if(i < argc - 1)
  586.                                 idfile = argv[++i];
  587.                         continue;
  588. 		}
  589. 		if(buf == "-d")
  590.                 {
  591.                         if(i < argc - 1)
  592.                                 distance_analysis_matrix = argv[++i];
  593.                         continue;
  594.                 }
  595. */
  596. 		if(buf == "-ql")
  597.                 {
  598.                         if(i < argc - 1)
  599.                                 q_idlist = argv[++i];
  600.                         continue;
  601.                 }
  602. /*
  603. 		if(buf == "-imgm")
  604.                 {
  605.                         if(i < argc - 1)
  606.                                 bg_for_img = argv[++i];
  607.                         continue;
  608.                 }
  609. 		if(buf == "-magic")
  610.                 {
  611.                         if(i < argc - 1)
  612.                                 magic = argv[++i];
  613.                         continue;
  614.                 }
  615. 		if(buf == "-track")
  616.                 {
  617.                         if(i < argc - 1)
  618.                                 trackbg = argv[++i];
  619.                         continue;
  620.                 }*/
  621.  
  622. 		else if(buf == "-p")
  623. 		{
  624. 			if(i < argc - 1)
  625. 				promfile = argv[++i];
  626. 			BIG_SEQ = false;
  627. 			continue;
  628. 		}
  629. /*
  630. 		else if(buf == "-P")
  631. 		{
  632. 			if(i < argc - 1)
  633.                                 promfile = argv[++i];
  634.                         BIG_SEQ = true;
  635. //			USE_N = true;
  636.                         continue;
  637. 		}
  639. 		else if(buf == "-chip")
  640. 		{
  641. 			i++;
  643. 			while(i < (argc))
  644. 			{
  645. 				buf = argv[i];
  647. 				if(buf[0] == '-')
  648. 				{
  649. 					i--;
  650. 					break;
  651. 				}
  652. 				else
  653. 				{
  654. 					chip_files.push_back(buf);
  655. 					i++;
  656. 				}
  657. 			}
  658. 							
  659. 			if(!chip_files.empty())
  660. 				CHIP = true;
  662. 			continue;
  663. 		}
  664. */
  665. 		else if(buf == "-m")
  666. 		{
  667. 			if(i < argc - 1)
  668. 				matrixfile = argv[++i];
  669. 			continue;
  670. 		}
  671. 		#ifdef BOUND_OLIGOS
  672. 		else if(buf == "-bo")
  673.                 {
  674.                         if(i < argc - 1)
  675.                                 bound_oligos_file = argv[++i];
  676.  
  677.                         continue;
  678.                 }
  679. 		#endif
  680.  
  681. /*		else if(buf == "-w2")
  682. 		{
  683. 			if(i < argc - 1)
  684. 				w2file = argv[++i];
  686. 			continue;
  687. 		}*/
  688. 		else if(buf == "-l")
  689.                 {
  690.                         if(i < argc - 1)
  691.                                 fasta_matrix_file = argv[++i];
  692.  
  693.                         continue;
  694.                 }
  695. 		else if(buf == "-ui")
  696.                 {
  697.                         if(i < argc - 1)
  698.                                 index_file = argv[++i];
  699.  
  700.                         continue;
  701.                 }
  702. 		else if(buf == "-M")
  703. 		{
  704. 			if(i < argc - 1)
  705. 				usethismatrix = argv[++i];
  706. 			continue;
  707. 		}
  708. 		else if(buf == "-N")
  709.                 {
  710.                         if(i < argc - 1)
  711.                                 usethismatrixname = argv[++i];
  712.                         continue;
  713.                 }
  714. /*
  715. 		else if(buf == "-L")
  716. 		{	
  717. 			if(i < argc - 1)
  718. 				matrixlist = argv[++i];
  719. 			continue;
  720. 		}*/
  721.  
  722. /*		else if(buf == "-start")
  723.                 {
  724.                         if(i < argc - 1)
  725. 			{
  726.                                  SEQ_EDGE_START = atoi(argv[++i]);
  727. 				 MIN_SEQ_LENGTH = SEQ_EDGE_START; 
  728. 			}
  729.                         continue;
  730.                 }
  731. 		else if(buf == "-end")
  732.                 {
  733.                         if(i < argc - 1)
  734. 			{
  735.                                  SEQ_EDGE_END = atoi(argv[++i]);
  736. 				 MIN_SEQ_LENGTH = SEQ_EDGE_END;
  737. 			}
  738.                         continue;
  739.                 }
  740. 		else if(buf == "-min")
  741.                 {
  742.                         if(i < argc - 1)
  743.                                  MIN_SEQ_LENGTH = atoi(argv[++i]);
  744.                         continue;
  745.                 }*/
  746. 		else if(buf == "-split")
  747. 		{
  748. 			if(i < argc - 2)
  749. 			{
  750. 				SPLIT[0] = atoi(argv[++i]);
  751. 				SPLIT[1] = atoi(argv[++i]);
  752. 			}
  753. 			continue;
  754. 		}
  755. /*		else if(buf == "-w")
  756. 		{
  757. 			if(i < argc - 1)
  758. 				BIG_WINDOW = atoi(argv[++i]);
  759. 			continue;
  760. 		}
  761. 		else if(buf == "-step")
  762. 		{
  763. 			if(i< argc - 1)
  764. 				BIG_STEP = atoi(argv[++i]);
  765. 			continue;
  766. 		}
  767. 		else if(buf == "-probe")
  768.                 {
  769.                         if(i< argc - 1)
  770.                                 PROBE = atoi(argv[++i]);
  772.                         continue;
  773.                 }
  774. 		else if(buf == "-tss")
  775.                 {
  776.                         if(i< argc - 1)
  777.                                 TSS_POS = -atoi(argv[++i]);
  778.                         continue;
  779.                 }
  780. 		else if(buf == "-cut")
  781.                 {
  782. 			istringstream str;
  784.                         if(i< argc - 1)
  785.                                 str.str(argv[++i]);
  787. 			str >> TRACK_CUTOFF;
  789.                         continue;
  790.                 }
  791. 		else if(buf == "-chip_cutoff")
  792. 		{
  793. 			istringstream str;
  795. 			if(i < argc - 1)
  796. 				str.str(argv[++i]);
  798. 			str >> CHIP_CUTOFF;
  799. 		}
  800. 		else if(buf == "-d_cutoff")
  801.                 {
  802.                         istringstream str;
  804.                         if(i < argc - 1)
  805.                                 str.str(argv[++i]);
  807.                         str >> DISTANCE_CUTOFF;
  808.                 }
  809. 		else if(buf == "-v")
  810. 		{
  811. 			VERBOSE = true;
  812. 			continue;
  813. 		}*/
  814. 		else if(buf == "-bi")
  815. 		{
  816. 			BUILD_INDEX = true;
  817. 			continue;
  818. 		}
  819. /*		else if(buf == "-drawbg")
  820.                 {
  821.                         DRAW_BACKGROUND = true;
  822.                         continue;
  823.                 }
  824. 		else if(buf == "-noheader")
  825.                 {
  826.                         NO_HEADER = true;
  827.                         continue;
  828.                 }
  829. 		else if(buf == "-cv")
  830. 		{
  831. 			SHOW_CONSERVATION_VECTOR = true;
  832. 			continue;
  833. 		}
  834. 		else if(buf == "-img")
  835.                 {
  836.                         IMG_OUTPUT = true;
  837.                         continue;
  838.                 }
  839. 		else if(buf == &
声明：本文内容由网友自发贡献，不代表【wpsshop博客】立场，版权归原作者所有，本站不承担相应法律责任。如您发现有侵权的内容，请联系我们。转载请注明出处：https://www.wpsshop.cn/w/繁依Fanyi0/article/detail/353105
推荐阅读
相关标签

Copyright © 2003-2013 www.wpsshop.cn 版权所有，并保留所有权利。

  

闽ICP备14008679号