peakrank_advanced.cpp

来自「MS-Clustering is designed to rapidly clu」· C++ 代码 · 共 1,022 行 · 第 1/3 页

CPP
1,022 行
字号
123
		sample.add_real_feature(r_idx+amino_acids[cut_idx],cut_prop);
	r_idx+=num_aas;


	// Add sepcial C-terminal features
	const int c_cut_dis = length - cut_idx;
	if (c_cut_dis<=5 && amino_acids[length-1]==LysIdx)
	{
		if (c_aa_counts[LeuIdx]>0 &&
			(c_aa_counts[AspIdx]+c_aa_counts[GluIdx]>0 &&
			c_aa_counts[LeuIdx] + c_aa_counts[GluIdx] + +c_aa_counts[AspIdx] +
			c_aa_counts[LysIdx] == c_cut_dis) )
			sample.add_real_feature(r_idx,c_cut_dis);
		r_idx++;

		if (c_aa_counts[LeuIdx]>0 && c_aa_counts[AlaIdx]>0 &&
			c_aa_counts[LeuIdx] + c_aa_counts[AlaIdx] + c_aa_counts[LysIdx] == c_cut_dis)
			sample.add_real_feature(r_idx,c_cut_dis);
		r_idx++;

		if (c_cut_dis>=3 && c_aa_counts[LeuIdx] + c_aa_counts[LysIdx] + c_aa_counts[GluIdx] + 
			c_aa_counts[AspIdx] == c_cut_dis-1)
			sample.add_real_feature(r_idx,c_cut_dis);
		r_idx++;
	}
	else
		r_idx+=3;

	if (amino_acids[length-1]==ArgIdx && cut_idx<length-1)
	{
		if (c_aa_counts[LeuIdx] + c_aa_counts[ArgIdx] + c_aa_counts[GluIdx] + 
			c_aa_counts[AspIdx] == c_cut_dis)
		sample.add_real_feature(r_idx,c_cut_dis);
		
	}
	r_idx++;
}



void push_back_all_RHK_pairs(vector<string>& real_feature_names, string prefix_label)
{
	int n;
	for (n=0; n<num_RKH_combos; n++)
	{
		int c;
		for (c=0; c<num_RKH_combos; c++)
		{
			string label = prefix_label + " " + combos[n] + "/" + combos[c];
			real_feature_names.push_back(label);
		}
	}
}


void PeakRankModel::set_advanced_feature_names()
{
	const int num_aas = this->get_num_model_aas();
	int i;

	real_feature_names.clear();
	real_feature_stage_idxs.clear();

	real_feature_names.push_back("N RKH LEVEL");
	real_feature_names.push_back("C RKH LEVEL");

	real_feature_names.push_back("PROP PEAK MASS IN VIS RANGE"); 
	push_back_all_RHK_pairs(real_feature_names,"PROP PEAK MASS IN VIS RANGE");

	real_feature_names.push_back("DIS FROM MIN");
	push_back_all_RHK_pairs(real_feature_names,"DIS FROM MIN");

	
	real_feature_names.push_back("DIS FROM MAX");
	push_back_all_RHK_pairs(real_feature_names,"DIS FROM MAX");

	real_feature_names.push_back("CUT POS");
	push_back_all_RHK_pairs(real_feature_names,"CUT POS");

	real_feature_names.push_back("DIS MIN<75, CUT POS");
	real_feature_names.push_back("DIS MIN<150, CUT POS");
	real_feature_names.push_back("DIS MIN<250, CUT POS");
	real_feature_names.push_back("DIS MIN<=400, CUT POS");

	real_feature_names.push_back("DIS MAX<75, CUT POS");
	real_feature_names.push_back("DIS MAX<150, CUT POS");
	real_feature_names.push_back("DIS MAX<250, CUT POS");
	real_feature_names.push_back("DIS MAX<=400, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("#N-SIDE " + model_aa_labels[i] + "(without flanking 3)");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("#C-SIDE " + model_aa_labels[i] + "(without flanking 3)");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("#N-SIDE " + model_aa_labels[i]);

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("#C-SIDE " + model_aa_labels[i]);

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -1 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -2 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -3 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +1 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +2 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +3 = " + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
	{
		int j;
		for (j=0; j<num_aas; j++)
			real_feature_names.push_back("CUT IS " + model_aa_labels[i] + "-" + model_aa_labels[j] + ", CUT POS");
	}

	for (i=0; i<num_aas; i++)
	{
		int j;
		for (j=0; j<num_aas; j++)
			real_feature_names.push_back("CUT IS " + model_aa_labels[i] + "-" + model_aa_labels[j] + ", PEAK PROP");
	}


	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -1 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -2 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -3 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +1 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +2 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +3 = " + model_aa_labels[i] + ", RHK LEVEL");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -1 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -2 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT -3 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +1 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +2 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back("CUT +3 = " + model_aa_labels[i] + ", PEAK PROP");

	for (i=0; i<num_aas; i++)
	{
		int j;
		for (j=0; j<num_aas; j++)
			real_feature_names.push_back(model_aa_labels[i] + model_aa_labels[j] + " BEFORE CUT, CUT POS");
	}

	for (i=0; i<num_aas; i++)
	{
		int j;
		for (j=0; j<num_aas; j++)
			real_feature_names.push_back(model_aa_labels[i] + model_aa_labels[j] + " AFTER CUT, CUT POS");
	}


	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXXXK, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|"+ model_aa_labels[i] + "XXXXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "|X"+ model_aa_labels[i] + "XXXXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXXR, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( model_aa_labels[i] + "|XXXXR, CUT POS");


	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "KXXXX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXXX" + model_aa_labels[i] + "|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXXX" + model_aa_labels[i] + "X|, CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXX|" + model_aa_labels[i] + ", CUT POS");

	for (i=0; i<num_aas; i++)
		real_feature_names.push_back( "RXXXX|" + model_aa_labels[i] + ", CUT POS");

	real_feature_names.push_back("|[LD/EK] - permuted");
	real_feature_names.push_back("|[LAK] - permuted");
	real_feature_names.push_back("|[LDEK] - permuted with one X");
	real_feature_names.push_back("|[LDER] - permuted");
}
123

peakrank_advanced.cpp - 源码说明

本页面展示了「MS-Clustering is designed to rapidly cluster large MS/MS datasets. The program merges similar spectr」中的 peakrank_advanced.cpp 源码文件，采用 C++ 编程语言编写，共 1,022 行代码。您可以在线阅读完整代码内容，也可以返回资源详情页下载完整源码包进行本地学习和开发。

虫虫开发者社区收录了大量与质谱分析相关的技术资源，包括源代码、技术文档、电路图等，是电子工程师和嵌入式开发者的专业学习平台。

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?