yamcot 1.0.0__cp311-cp311-macosx_11_0_arm64.whl

yamcot/__init__.py

@@ -0,0 +1,46 @@
+"""
+unimotifcomparator
+==================
+
+This package provides a modular and extensible framework for de‑novo motif
+discovery, motif comparison and performance evaluation in ChIP‑seq like
+applications.  The design emphasises object oriented abstractions and clear
+interfaces so that new motif discovery tools, motif comparison strategies
+and evaluation metrics can be added with minimal effort.  All code is
+intended for educational and fundamental research use on open, anonymised
+datasets only and must not be applied to sensitive or identifiable data.
+
+The top level modules expose the following key components:
+
+``io``
+    Functions for reading and writing biological sequences in FASTA format
+    and motifs in MEME format.
+
+``models``
+    Motif model classes.  A generic :class:`MotifModel` describes the common
+    behaviour of sequence motifs, while concrete subclasses such as
+    :class:`PWMMotif` implement specific scoring logic.
+
+``discovery``
+    Base classes and concrete implementations of motif discovery tools.
+
+``comparison``
+    Base classes and concrete implementations of motif comparison metrics.
+
+``evaluation``
+    Classes for computing ROC/PR curves and their associated summary
+    statistics.
+
+``pipeline``
+    High level orchestrators that perform bootstrapping, odd/even
+    cross‑validation, motif comparison and final motif selection.
+
+``cli``
+    An example command line interface exposing the pipeline to end users.
+
+This modular structure makes it straightforward to integrate alternative
+motif models (e.g. BaMM or SiteGA), comparison methods (e.g. Jaccard
+index, overlap coefficient or recognition‑function correlation) and
+performance metrics.  See individual modules for detailed API
+documentation.
+"""

yamcot/_core/__init__.py

@@ -0,0 +1,17 @@
+# Import the C++ extension functionality
+try:
+    # Try to import the compiled C++ extension directly
+    from . import _core
+
+    run_motali_cpp = _core.run_motali_cpp
+except (ImportError, AttributeError) as e:
+    _import_error = e
+    _core = None
+
+    # Fallback when the compiled extension is not available
+    def run_motali_cpp(*args, **kwargs):
+        raise ImportError(
+            "The C++ extension '_core' is not installed or could not be loaded. "
+            "Please ensure the package was built correctly. "
+            f"Original error: {_import_error}"
+        )

yamcot/_core/bindings.cpp

@@ -0,0 +1,28 @@
+#include <nanobind/nanobind.h>
+#include <nanobind/stl/string.h>
+
+#include "core_functions.h"
+
+namespace nb = nanobind;
+
+NB_MODULE(_core, m) {
+    m.doc() = "C++ backend for yamcot";
+    
+    m.def("run_motali_cpp", &run_motali_cpp,
+          nb::arg("file_fasta"),
+          nb::arg("type_model_1"), 
+          nb::arg("type_model_2"),
+          nb::arg("file_model_1"),
+          nb::arg("file_model_2"), 
+          nb::arg("file_table_1"),
+          nb::arg("file_table_2"),
+          nb::arg("shift"),
+          nb::arg("threshold"),
+          nb::arg("file_hist"),
+          nb::arg("yes_out_hist"),
+          nb::arg("file_prc"),
+          nb::arg("yes_out_prc"),
+          nb::arg("file_short_over"),
+          nb::arg("file_short_all"),
+          nb::arg("file_sta_long"));
+}

yamcot/_core/core_functions.h

@@ -0,0 +1,29 @@
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Function declaration for the main computational function
+int run_motali_cpp(
+    const char* file_fasta,
+    const char* type_model_1,
+    const char* type_model_2,
+    const char* file_model_1,
+    const char* file_model_2,
+    const char* file_table_1,
+    const char* file_table_2,
+    int shift,
+    double pvalue,
+    const char* file_hist,
+    int yes_out_hist,
+    const char* file_prc,
+    int yes_out_prc,
+    const char* file_short_over,
+    const char* file_short_all,
+    const char* file_sta_long
+);
+
+#ifdef __cplusplus
+}
+#endif

yamcot/_core/fasta_to_plain.h

@@ -0,0 +1,182 @@
+char TransStr(char x)
+{
+  int c=int(x);
+  if(c<97) x=char(c+32);	
+   return x;
+}
+int fasta_to_plain_genome(char* file_in_fasta, int motif_len_min, int& all_pos, int &nseq, int &len)
+{
+	char l[SEQLEN], d[SEQLEN], head[400];
+	int fl = 0;
+	FILE* in;
+
+	if ((in = fopen(file_in_fasta, "rt")) == NULL)
+	{
+		printf("Input file %s can't be opened\n", file_in_fasta);
+		return -1;
+	}
+	char symbol = fgetc(in);
+	rewind(in);
+	all_pos = nseq = len = 0;
+	while (nseq >= 0)
+	{
+		if (fgets(l, sizeof(l), in) == NULL) fl = -1;
+		if (*l == '\n' && fl != -1)continue;
+		if (((*l == symbol) || (fl == -1)) && (fl != 0))
+		{
+			int lenx = (int)strlen(d);
+			if (lenx > len)len = lenx;
+			all_pos += (lenx-motif_len_min);
+			nseq++;
+			//	int check = CheckStr(file, d, n, 1, outlog);
+			//if (check != -1)			
+			if (fl == -1)
+			{
+				fclose(in);				
+				break;
+			}			
+		}
+		if (*l == symbol)
+		{
+			memset(head, 0, sizeof(head));
+			DelChar(l, '\n');
+			strcpy(head, l);
+			fl = 0; continue;
+		}
+		if (fl == 0)
+		{
+			memset(d, 0, sizeof(d));
+			DelChar(l, '\n');
+			strcpy(d, l);
+			fl = 1; continue;
+		}
+		if (strlen(d) + strlen(l) > sizeof(d))
+		{
+			printf("Size is large...");
+			printf("l:%s\nstrlen(l):%zu\n", l, strlen(l));
+			printf("d:%s\nstrlen(d):%zu\n", d, strlen(d));
+			exit(1);
+		}
+		DelChar(l, '\n');
+		strcat(d, l);
+	}
+	return 1;
+}
+int fasta_to_plain0(char *file_in_fasta, int &length_fasta_max, int &nseq_fasta)
+{
+	char head[200];		
+	FILE *in;
+
+	if((in=fopen(file_in_fasta,"rt"))==NULL)
+	{
+		printf("Input file %s can't be opened\n",file_in_fasta);
+		return -1;
+	}
+	char c, symbol = '>';	
+	nseq_fasta=0;
+	int len=0;
+	//double sum_len=0;
+	int fl=1;
+	length_fasta_max=0;
+	do  
+	{
+		c=getc(in);
+		if(c==EOF)fl=-1;
+		if(c==symbol || fl==-1)
+		{
+			if(nseq_fasta>0)
+			{
+				if(len>length_fasta_max)length_fasta_max=len;
+				if(len>SEQLEN)
+				{
+					printf("Sequence N %d too long... %d nt\n",nseq_fasta,len);
+					return -1;
+				}
+			}			
+			if(fl!=-1)
+			{
+				nseq_fasta++;				
+				len=0;
+			}
+			if(fl==1)
+			{
+				fgets(head,sizeof(head),in);			
+				continue;
+			}
+		}
+		if(c=='\n')continue;	
+		if(c=='\t')continue;	
+		if(c==' ')continue;	
+		if(c=='\r')continue;	
+		len++;
+	}
+	while(fl==1);
+	fclose(in);
+	return 1;
+}
+int fasta_to_plain1(char *file_in_fasta, int length_fasta_max, int nseq_fasta, char ***seq, int *peak_len)
+{
+	int fl=1, n=0, len=0;
+	char head[200];		
+	char c, symbol = '>';
+	char alfavit4[]="acgtnACGTN";
+	FILE *in;
+	if((in=fopen(file_in_fasta,"rt"))==NULL)
+	{
+		printf("Input file %s can't be opened\n",file_in_fasta);
+		return -1;
+	}	
+	do  
+	{
+		c=getc(in);
+		if(c==EOF)
+		{
+			fl=-1;
+		}
+		if(c==symbol || fl==-1)
+		{		
+			if(len>0)
+			{	
+				peak_len[n]=len;
+				seq[0][n][len]='\0';
+				strncpy(seq[1][n],seq[0][n],len);
+				seq[1][n][len]='\0';
+				ComplStr(seq[1][n]);								
+				{
+					if(fl!=-1)
+					{
+						n++;						
+						len=0;
+					}
+				}
+			}
+			else
+			{
+				if(n>0 && fl!=-1)
+				{
+					printf("Peak length error! peak %d\n",n+1);
+					return -1;
+				}
+			}
+			if(fl==-1)break;
+			if(fl==1)
+			{
+				fgets(head,sizeof(head),in);			
+				continue;
+			}
+		}
+		if(c=='\n')continue;	
+		if(c=='\t')continue;	
+		if(c==' ')continue;	
+		if(c=='\r')continue;	
+		if(strchr(alfavit4,c)!=NULL)
+		{
+			c=TransStr(c);
+			seq[0][n][len++]=c;
+		}
+		else seq[0][n][len++]='n';
+	}
+	while(fl==1);
+	fclose(in);	
+	return 1;
+}