Context Navigation

source: cpp/frams/genetics/fB/fB_oper.cpp @ 912

Last change on this file since 912 was 896, checked in by Maciej Komosinski, 5 years ago
Replaced #defined macros for popular random-related operations with functions
File size: 14.8 KB

Rev	Line
[797]	1	// This file is a part of Framsticks SDK. http://www.framsticks.com/
[853]	2	// Copyright (C) 1999-2019 Maciej Komosinski and Szymon Ulatowski.
[797]	3	// See LICENSE.txt for details.
	4
[780]	5	#include <frams/util/sstring.h>
	6	#include <vector>
	7	#include <frams/param/param.h>
	8	#include "fB_conv.h"
	9	#include "fB_general.h"
	10	#include "fB_oper.h"
[797]	11	#include "../fH/fH_oper.h"
[780]	12
	13	#define FIELDSTRUCT Geno_fB
	14
	15	static ParamEntry GENOfBparam_tab[] =
	16	{
[797]	17	{ "Genetics: fB", 3, FB_MUT_COUNT + FB_XOVER_COUNT, },
[780]	18	{ "Genetics: fB: Mutation", },
	19	{ "Genetics: fB: Crossover", },
	20	{ "fB_mut_substitution", 1, 0, "Substitution", "f 0 1 0.6", FIELD(mutationprobs[FB_SUBSTITUTION]), "Probability of mutation by changing single random letter in genotype", },
[802]	21	{ "fB_mut_insertion", 1, 0, "Insertion", "f 0 1 0.095", FIELD(mutationprobs[FB_INSERTION]), "Probability of mutation by inserting characters in random place of genotype", },
	22	{ "fB_mut_nclassins", 1, 0, "Insertion of neuron class definition", "f 0 1 0.005", FIELD(mutationprobs[FB_NCLASSINS]), "Probability of mutation by inserting neuron class definition in random place of genotype", },
[780]	23	{ "fB_mut_deletion", 1, 0, "Deletion", "f 0 1 0.1", FIELD(mutationprobs[FB_DELETION]), "Probability of mutation by deleting random characters in genotype", },
[802]	24	{ "fB_mut_duplication", 1, 0, "Duplication", "f 0 1 0.0", FIELD(mutationprobs[FB_DUPLICATION]), "Probability of mutation by copying single gene of genotype and appending it to the beginning of this genotype", },
[780]	25	{ "fB_mut_translocation", 1, 0, "Translocation", "f 0 1 0.15", FIELD(mutationprobs[FB_TRANSLOCATION]), "Probability of mutation by replacing two substrings in genotype", },
[802]	26	{ "fB_cross_gene_transfer", 2, 0, "Horizontal gene transfer", "f 0 1 0.0", FIELD(crossoverprobs[FB_GENE_TRANSFER]), "Probability of crossing over by transferring single genes from both parents to beginning of each other", },
	27	{ "fB_cross_crossover", 2, 0, "Crossing over", "f 0 1 1.0", FIELD(crossoverprobs[FB_CROSSING_OVER]), "Probability of crossing over by random distribution of genes from both parents to both children", },
[780]	28	{ 0, },
	29	};
	30
	31	#undef FIELDSTRUCT
	32
	33	Geno_fB::Geno_fB()
	34	{
	35	par.setParamTab(GENOfBparam_tab);
	36	par.select(this);
	37	par.setDefault();
	38	supported_format = 'B';
	39	}
	40
[821]	41	bool Geno_fB::hasStick(const SString &genotype)
[780]	42	{
	43	for (int i = 0; i < fB_GenoHelpers::geneCount(genotype); i++)
	44	{
	45	int start, end;
	46	SString gene = fB_GenoHelpers::getGene(i, genotype, start, end);
	47	int endoffset = 0;
	48	if (gene.indexOf("zz", 0) != -1) endoffset = 2;
	49	if (gene.len() - endoffset < 3)
	50	{
	51	return true; // genes with length < 3 are always sticks
	52	}
	53	else if (gene[2] >= 'a' && gene[2] <= 'i')
	54	{
	55	return true; // gene within this range is stick
	56	}
	57	}
	58	return false;
	59	}
	60
	61	int Geno_fB::checkValidity(const char geno, const char genoname)
	62	{
	63	// load genotype
	64	SString genotype(geno);
	65	SString line;
	66	int pos = 0;
	67	// if there is no genotype to load, then return error
	68	if (!genotype.getNextToken(pos, line, '\n'))
	69	{
	70	return pos + 1;
	71	}
	72	// extract dimensions
	73	int dims = 0;
	74	if (!ExtValue::parseInt(line.c_str(), dims, true, false))
	75	{
	76	return 1;
	77	}
	78	// extract next token in order to check if next line starts with "aa"
	79	int genstart = genotype.indexOf("aa", 0);
	80	if (genstart != pos)
	81	{
	82	return pos + 1;
	83	}
	84	// check if rest of characters are lowercase
	85	for (int i = genstart; i < genotype.len(); i++)
	86	{
	87	if (!islower(genotype[i]))
	88	{
[797]	89	if (genotype[i] == '"')
	90	{
	91	SString neuclassdef;
	92	int nextid = i + 1;
	93	if (!genotype.getNextToken(nextid, neuclassdef, '"'))
	94	{
	95	return i + 1;
	96	}
	97	Neuro *neu = new Neuro();
	98	neu->setDetails(neuclassdef);
	99
	100	bool isclass = neu->getClass() ? true : false;
	101	delete neu;
	102	if (!isclass)
	103	{
	104	return i + 1;
	105	}
	106	i = nextid;
	107	}
	108	else
	109	{
	110	return i + 1;
	111	}
[780]	112	}
	113	}
	114	if (!hasStick(genotype))
	115	{
	116	return 1;
	117	}
	118	return GENOPER_OK;
	119	}
	120
	121	int Geno_fB::validate(char &geno, const char genoname)
	122	{
	123	// load genotype
	124	SString genotype(geno);
	125	SString strdims;
	126	int pos = 0;
	127	if (!genotype.getNextToken(pos, strdims, '\n'))
	128	{
	129	return GENOPER_OPFAIL;
	130	}
	131	// parse dimension
	132	int dims = 0;
	133	if (!ExtValue::parseInt(strdims.c_str(), dims, true, false))
	134	{
	135	return GENOPER_OPFAIL;
	136	}
	137	SString line;
	138	bool fix = false;
	139	int genstart = genotype.indexOf("aa", 0);
	140	// if there is no "aa" codon in the beginning of a genotype, then add it
	141	if (genstart != pos)
	142	{
	143	genotype = strdims + "\naa" + genotype.substr(pos);
	144	fix = true;
	145	}
	146	for (int i = pos; i < genotype.len(); i++)
	147	{
	148	// if character is not alphabetic - error
	149	if (!isalpha(genotype[i]))
	150	{
[797]	151	if (genotype[i] == '"')
	152	{
	153	SString neuclassdef;
	154	int nextid = i + 1;
	155	if (!genotype.getNextToken(nextid, neuclassdef, '"'))
	156	{
	157	return i + 1;
	158	}
	159	Neuro *neu = new Neuro();
	160	neu->setDetails(neuclassdef);
	161
	162	bool isclass = neu->getClass() ? true : false;
	163	delete neu;
	164	if (!isclass)
	165	{
	166	return i + 1;
	167	}
	168	i = nextid;
	169	}
	170	else
	171	{
	172	return GENOPER_OPFAIL;
	173	}
[780]	174	}
	175	// if character is uppercase, then convert it to lowercase
[797]	176	else if (isupper(genotype[i]))
[780]	177	{
	178	genotype.directWrite()[i] = tolower(genotype[i]);
	179	fix = true;
	180	}
	181	}
	182	// if the genotype does not contain any stick - add it
	183	if (!hasStick(genotype))
	184	{
	185	genotype = SString("aaazz") + genotype;
	186	}
	187	// if there were any changes - save them
	188	if (fix)
	189	{
	190	free(geno);
	191	geno = strdup(genotype.c_str());
	192	}
	193	return GENOPER_OK;
	194	}
	195
[821]	196	SString Geno_fB::detokenizeSequence(std::list<SString> *tokenlist)
[797]	197	{
	198	SString res = "";
[821]	199	for (std::list<SString>::iterator it = tokenlist->begin(); it != tokenlist->end(); it++)
[797]	200	{
	201	res += (*it);
	202	}
	203	return res;
	204	}
	205
[821]	206	std::list<SString> Geno_fB::tokenizeSequence(const SString &genotype)
[797]	207	{
	208	std::list<SString> res;
	209	int i = 0;
	210	while (i < genotype.len())
	211	{
	212	// if character is not alphabetic - error
	213	if (isalpha(genotype[i]))
	214	{
	215	SString el = "";
	216	el += genotype[i];
	217	res.push_back(el);
	218	i++;
	219	}
	220	else
	221	{
	222	SString neuclassdef;
	223	i++;
	224	genotype.getNextToken(i, neuclassdef, '"');
	225	SString ndef = "\"";
	226	ndef += neuclassdef;
	227	ndef += "\"";
	228	res.push_back(ndef);
	229	}
	230	}
	231	return res;
	232	}
	233
[780]	234	int Geno_fB::mutate(char *&geno, float &chg, int &method)
	235	{
	236	SString genotype(geno);
	237	SString strdims;
	238	int pos = 0;
	239	genotype.getNextToken(pos, strdims, '\n');
	240	SString line;
	241	genotype.getNextToken(pos, line, '\n');
	242	method = roulette(mutationprobs, FB_MUT_COUNT);
	243	switch (method)
	244	{
	245	case FB_SUBSTITUTION:
	246	{
[797]	247	std::list<SString> tokenized = tokenizeSequence(line);
[896]	248	int rndid = rndUint(tokenized.size()); // select random letter from genotype
[780]	249	// increment/decrement character - when overflow happens, this method
	250	// uses reflect method
[797]	251	std::list<SString>::iterator it = tokenized.begin();
	252	std::advance(it, rndid);
	253	SString t = (*it);
	254	if ((*it).len() == 1)
[780]	255	{
[896]	256	if (rndUint(2) == 0)
[797]	257	{
	258	if ((it)[0] == 'a') (it).directWrite()[0] = 'b';
	259	else (it).directWrite()[0] = (it)[0] - 1;
	260	}
	261	else
	262	{
	263	if ((it)[0] == 'z') (it).directWrite()[0] = 'y';
	264	else (it).directWrite()[0] = (it)[0] + 1;
	265	}
	266	chg = 1.0 / line.len();
[780]	267	}
	268	else
	269	{
[797]	270	// first method needs to extract quotes
	271	SString def = (*it);
	272	def = def.substr(1, def.len() - 2);
	273	Geno_fH::mutateNeuronProperties(def);
	274	SString res = "\"";
	275	res += def;
	276	res += "\"";
	277	(*it) = res;
	278	chg = (double)def.len() / line.len();
[780]	279	}
[821]	280	line = detokenizeSequence(&tokenized);
[780]	281	break;
	282	}
[802]	283	case FB_NCLASSINS:
	284	{
	285	std::list<SString> tokenized = tokenizeSequence(line);
	286	std::list<SString>::iterator it = tokenized.begin();
[896]	287	int rndid = rndUint(tokenized.size()); // select random insertion point
[802]	288	std::advance(it, rndid);
	289	NeuroClass *cls = getRandomNeuroClass();
	290	if (cls)
	291	{
	292	SString classdef = cls->getName();
	293	Geno_fH::mutateNeuronProperties(classdef);
	294	SString res = "\"";
	295	res += classdef;
	296	res += "\"";
	297	tokenized.insert(it, res);
	298	chg = (double)classdef.len() / line.len();
[821]	299	line = detokenizeSequence(&tokenized);
[802]	300	break;
	301	}
	302	}
[853]	303	[[fallthrough]];
[780]	304	case FB_INSERTION:
	305	{
	306	chg = 1.0 / line.len();
[797]	307	std::list<SString> tokenized = tokenizeSequence(line);
[896]	308	int rndid = rndUint(tokenized.size()); // select random insertion point
[797]	309	std::list<SString>::iterator it = tokenized.begin();
	310	std::advance(it, rndid);
	311	SString letter = "a";
[896]	312	letter.directWrite()[0] = 'a' + rndUint(26);
[797]	313	tokenized.insert(it, letter);
[821]	314	line = detokenizeSequence(&tokenized);
[780]	315	break;
	316	}
	317	case FB_DELETION:
	318	{
	319	chg = 1.0 / line.len();
[797]	320	std::list<SString> tokenized = tokenizeSequence(line);
	321	std::list<SString>::iterator it = tokenized.begin();
[896]	322	int rndid = rndUint(tokenized.size()); // select random deletion point
[797]	323	std::advance(it, rndid);
	324	tokenized.erase(it);
[821]	325	line = detokenizeSequence(&tokenized);
[780]	326	break;
	327	}
	328	case FB_DUPLICATION:
	329	{
[896]	330	int rndgene = rndUint(fB_GenoHelpers::geneCount(line));
[780]	331	int start, end;
	332	SString gene = fB_GenoHelpers::getGene(rndgene, line, start, end);
	333	if (gene.indexOf("zz", 0) == -1) gene += "zz";
	334	chg = (float)gene.len() / line.len();
	335	line = gene + line;
	336	break;
	337	}
	338	case FB_TRANSLOCATION:
	339	{
[797]	340	std::list<SString> tokenized = tokenizeSequence(line);
	341	std::vector<unsigned int> cuts(4);
[780]	342	for (int i = 0; i < 4; i++)
	343	{
[896]	344	cuts[i] = rndUint(tokenized.size());
[780]	345	}
	346	std::sort(cuts.begin(), cuts.end());
[797]	347	std::vector<std::list<SString>::iterator> iters(4);
	348	for (int i = 0; i < 4; i++)
	349	{
	350	iters[i] = tokenized.begin();
	351	std::advance(iters[i], cuts[i]);
	352	}
	353
	354	std::list<SString> res;
	355	res.insert(res.end(), tokenized.begin(), iters[0]);
	356	res.insert(res.end(), iters[2], iters[3]);
	357	res.insert(res.end(), iters[1], iters[2]);
	358	res.insert(res.end(), iters[0], iters[1]);
	359	res.insert(res.end(), iters[3], tokenized.end());
	360
[853]	361	// SString first = line.substr(cuts[0], cuts[1] - cuts[0]);
	362	// SString second = line.substr(cuts[2], cuts[3] - cuts[2]);
	363	// SString result = line.substr(0, cuts[0]) + second +
	364	// line.substr(cuts[1], cuts[2] - cuts[1]) + first + line.substr(cuts[3]);
[821]	365	line = detokenizeSequence(&res);
[780]	366	chg = (float)(cuts[3] - cuts[2] + cuts[1] - cuts[0]) / line.len();
	367	break;
	368	}
	369	}
	370	SString result = strdims + "\n" + line;
	371	free(geno);
	372	geno = strdup(result.c_str());
	373	return GENOPER_OK;
	374	}
	375
	376	int Geno_fB::crossOver(char &g1, char &g2, float& chg1, float& chg2)
	377	{
	378	SString p1(g1);
	379	SString p2(g2);
	380
	381	int dims1 = 0, dims2 = 0;
	382	int pos = 0;
	383	SString strdims;
	384	p1.getNextToken(pos, strdims, '\n');
	385	ExtValue::parseInt(strdims.c_str(), dims1, true, false);
	386	SString parent1;
	387	p1.getNextToken(pos, parent1, '\n');
	388
	389	pos = 0;
	390	p2.getNextToken(pos, strdims, '\n');
	391	ExtValue::parseInt(strdims.c_str(), dims2, true, false);
	392
	393	if (dims1 != dims2)
	394	{
	395	return GENOPER_OPFAIL;
	396	}
	397
	398	SString parent2;
	399	p2.getNextToken(pos, parent2, '\n');
	400
	401	SString child1 = "";
	402	SString child2 = "";
	403
	404	switch (roulette(crossoverprobs, FB_XOVER_COUNT))
	405	{
	406	case FB_GENE_TRANSFER:
	407	{
	408	// get random gene from first parent
[896]	409	int choice = rndUint(fB_GenoHelpers::geneCount(parent1));
[780]	410	int start, end;
	411	SString gene = fB_GenoHelpers::getGene(choice, parent1, start, end);
	412	// add this gene to the beginning of the second parent genotype
	413	child2 = gene + parent2;
	414	chg2 = (float)parent2.len() / (float)child2.len();
	415	// do the same for second parent
[896]	416	choice = rndUint(fB_GenoHelpers::geneCount(parent2));
[780]	417	gene = fB_GenoHelpers::getGene(choice, parent2, start, end);
	418	child1 = gene + parent1;
	419	chg1 = (float)parent1.len() / (float)child1.len();
	420	break;
	421	}
[853]	422	// case FB_CROSSING_OVER:
	423	// {
	424	// // iterate through all genes of the first parent and assign them
	425	// // randomly to children
	426	// for (int i = 0; i < fB_GenoHelpers::geneCount(parent1); i++)
	427	// {
	428	// int start, end;
	429	// SString gene = fB_GenoHelpers::getGene(i, parent1, start, end);
[896]	430	// if (rndUint(2) == 0)
[853]	431	// {
	432	// child1 += gene;
	433	// chg1 += 1.0f;
	434	// }
	435	// else
	436	// {
	437	// child2 += gene;
	438	// }
	439	// }
	440	// chg1 /= fB_GenoHelpers::geneCount(parent1);
	441	//
	442	// // do the same with second parent
	443	// for (int i = 0; i < fB_GenoHelpers::geneCount(parent2); i++)
	444	// {
	445	// int start, end;
	446	// SString gene = fB_GenoHelpers::getGene(i, parent2, start, end);
[896]	447	// if (rndUint(2) == 0)
[853]	448	// {
	449	// child1 += gene;
	450	// }
	451	// else
	452	// {
	453	// child2 += gene;
	454	// chg2 += 1.0f;
	455	// }
	456	// }
	457	// chg2 /= fB_GenoHelpers::geneCount(parent2);
	458	// break;
	459	// }
[780]	460	case FB_CROSSING_OVER:
	461	{
[802]	462	// get maximal count of genes from both parents
	463	int maxgenecount = max(fB_GenoHelpers::geneCountNoNested(parent1),
[853]	464	fB_GenoHelpers::geneCountNoNested(parent2));
[802]	465
	466	// while there are genes in at least one genotype
	467	for (int i = 0; i < maxgenecount; i++)
[780]	468	{
[802]	469	SString to1 = "", to2 = "";
	470	int start = 0, end = 0;
	471
	472	// if both parents have genes available, then distribute them
	473	if (i < fB_GenoHelpers::geneCountNoNested(parent1) &&
[853]	474	i < fB_GenoHelpers::geneCountNoNested(parent2))
[780]	475	{
[896]	476	if (rndUint(2) == 0)
[802]	477	{
	478	to1 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
	479	to2 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
	480	chg1 += 1.0f;
	481	chg2 += 1.0f;
	482	}
	483	else
	484	{
	485	to1 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
	486	to2 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
	487	}
[780]	488	}
[802]	489	else if (i < fB_GenoHelpers::geneCountNoNested(parent1))
[780]	490	{
[896]	491	if (rndUint(2) == 0)
[802]	492	{
	493	to1 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
	494	chg1 += 1.0f;
	495	}
	496	else
	497	{
	498	to2 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
	499	}
[780]	500	}
[802]	501	else // if (i < fB_GenoHelpers::geneCountNoNested(parent2))
[780]	502	{
[896]	503	if (rndUint(2) == 0)
[802]	504	{
	505	to1 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
	506	}
	507	else
	508	{
	509	to2 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
	510	chg2 += 1.0f;
	511	}
[780]	512	}
[802]	513	child1 += to1;
	514	child2 += to2;
[780]	515	}
[802]	516
	517	chg1 /= fB_GenoHelpers::geneCountNoNested(parent1);
	518	chg2 /= fB_GenoHelpers::geneCountNoNested(parent2);
[780]	519	break;
	520	}
	521	}
	522
	523	free(g1);
	524	free(g2);
	525	if (child1.len() > 0 && child2.len() == 0)
	526	{
	527	child1 = strdims + "\n" + child1;
	528	g1 = strdup(child1.c_str());
	529	g2 = strdup("");
	530	}
	531	else if (child2.len() > 0 && child1.len() == 0)
	532	{
	533	child2 = strdims + "\n" + child2;
	534	g1 = strdup(child2.c_str());
	535	g2 = strdup("");
	536	}
	537	else
	538	{
	539	child1 = strdims + "\n" + child1;
	540	child2 = strdims + "\n" + child2;
	541	g1 = strdup(child1.c_str());
	542	g2 = strdup(child2.c_str());
	543	}
	544	return GENOPER_OK;
	545	}
	546
	547	uint32_t Geno_fB::style(const char *geno, int pos)
	548	{
	549	char ch = geno[pos];
	550	if (isdigit(ch))
	551	{
	552	while (pos > 0)
	553	{
	554	pos--;
	555	if (isdigit(geno[pos]) == 0)
	556	{
	557	return GENSTYLE_CS(0, GENSTYLE_INVALID);
	558	}
	559	}
	560	return GENSTYLE_RGBS(0, 0, 200, GENSTYLE_BOLD);
	561	}
	562	if (islower(ch) == 0)
	563	{
	564	return GENSTYLE_CS(0, GENSTYLE_INVALID);
	565	}
	566	uint32_t style = GENSTYLE_CS(GENCOLOR_TEXT, GENSTYLE_NONE);
	567	if (ch == 'a' && pos > 0 && (geno[pos - 1] == 'a' \|\| geno[pos - 1] == '\n'))
	568	{
	569	style = GENSTYLE_RGBS(0, 200, 0, GENSTYLE_BOLD);
	570	}
	571	else if (ch == 'z' && pos > 0 && geno[pos - 1] == 'z')
	572	{
	573	style = GENSTYLE_RGBS(200, 0, 0, GENSTYLE_BOLD);
	574	}
	575	return style;
	576	}

Note: See TracBrowser for help on using the repository browser.

Download in other formats: