codecreatede

final release version and usage added

README.md

@@ -6,4 +6,21 @@ go-whole-genome-estimates
 - the first genome will be taken as a reference. 
 - check the AlignmentGO package for defining your own block size. 
 
+```
+[gauravsablok@fedora]~/Desktop/codecreatede/whole-genome-alignment-estimate% \
+go run main.go -alignment ./samplefile/samplealignment.fasta
+The collinearity block for A as a base pattern and T as a mismatch is 1
+The collinearity block for A as a base pattern G as a mismatch is  1
+The collinearity block for A as a base pattern and C as a mismatch is 1
+The collinearity block for T as a base pattern and G as a mismatch is 0
+The collinearity block for T as a base pattern C as a mismatch is  0
+The collinearity block for T as a base pattern and A as a mismatch is 0
+The collinearity block for G as a base pattern and C as a mismatch is 1
+The collinearity block for G as a base pattern A as a mismatch is  0
+The collinearity block for G as a base pattern and T as a mismatch is 0
+The collinearity block for C as a base pattern and A as a mismatch is 0
+The collinearity block for C as a base pattern T as a mismatch is  0
+The collinearity block for C as a base pattern and G as a mismatch is 0
+```
+
 Gaurav Sablok

main.go

@@ -6,7 +6,6 @@ import (
 	"fmt"
 	"log"
 	"os"
-	"strconv"
 	"strings"
 )
 
@@ -20,13 +19,12 @@ It takes a whole genome aligned files and then estimates the linear block by usi
 You can pass as many genome as a reference genome and it will take the first genome as a reference. This only compares the
 single line block. In the alignmentGO package, you can find the specific for defining your own block.
 
- */
+*/
 
- func main () {
+func main() {
+	alignment := flag.String("alignment", "path to the alignment file", "file")
 
- alignment := flag.String("alignment", "path to the alignment file", "file")
-
- flag.Parse()
+	flag.Parse()
 
 	type alignmentIDStore struct {
 		id string
@@ -63,35 +61,115 @@ single line block. In the alignmentGO package, you can find the specific for def
 		}
 	}
 
-	counterA := 0
-	counterT := 0
-	counterG := 0
-	counterC := 0
+	counterAT := 0
+	counterAG := 0
+	counterAC := 0
+
+	for i := 0; i < len(sequenceSpec)-1; i++ {
+		for j := 0; j < len(sequenceSpec[0]); j++ {
+			if string(sequenceSpec[i][j]) == "A" && string(sequenceSpec[i+1][j]) == "T" {
+				counterAT++
+			}
+			if string(sequenceSpec[i][j]) == "A" && string(sequenceSpec[i+1][j]) == "C" {
+				counterAG++
+			}
+			if string(sequenceSpec[i][j]) == "A" && string(sequenceSpec[i+1][j]) == "G" {
+				counterAC++
+			}
+		}
+	}
+
+	counterTG := 0
+	counterTC := 0
+	counterTA := 0
 
 	for i := 0; i < len(sequenceSpec)-1; i++ {
 		for j := 0; j < len(sequenceSpec[0]); j++ {
-			if string(sequenceSpec[i][j]) == "A" && string(sequenceSpec[i+1][j]) == "T",
-			 || string(sequenceSpec[i+1][j] == "C" || string(sequenceSpec[i+1][j]) == "G" {
-					counterA++
-				}
-				if string(sequenceSpec[i][j]) == "T" && string(sequenceSpec[i+1][j]) == "C",
-					|| string(sequenceSpec[i+1][j]) == "G" || string(sequenceSpec[i+1][j]) == "A" {
-						counterT++
-					}
-					if string(sequenceSpec[i][j]) == "C" && string(sequenceSpec[i+1][j]) == "G",
-						|| string(sequenceSpec[i+1][j]) == "A" || string(sequenceSpec[i+1][j]) == "T" {
-							counterC++
-						}
-						if string(sequenceSpec[i][j]) == "G" && string(sequenceSpec[i+1][j]) == "A",
-							|| string(SequenceSpec[i+1][j]) == "T" || string(sequenceSpec[i+1][j]) == "C" {
-								counterG++
-							} else {
-								continue
-							}
+			if string(sequenceSpec[i][j]) == "T" && string(sequenceSpec[i+1][j]) == "G" {
+				counterTA++
+			}
+			if string(sequenceSpec[i][j]) == "T" && string(sequenceSpec[i+1][j]) == "C" {
+				counterTC++
+			}
+			if string(sequenceSpec[i][j]) == "T" && string(sequenceSpec[i+1][j]) == "A" {
+				counterTA++
+			}
 		}
 	}
-	fmt.Printf("The collinearity block of line width 1 for A as a base pattern is %d\n", counterA)
-	fmt.Printf("The collinearity block of line width 1 for T as a base pattern is %d\n", counterT)
-	fmt.Printf("The collinearity block of line width 1 for G as a base pattern is %d\n", counterG)
-	fmt.Printf("The collinearity block of line width 1 for C as a base pattern is %d\n", counterC)
+
+	counterGC := 0
+	counterGA := 0
+	counterGT := 0
+
+	for i := 0; i < len(sequenceSpec)-1; i++ {
+		for j := 0; j < len(sequenceSpec[0]); j++ {
+			if string(sequenceSpec[i][j]) == "G" && string(sequenceSpec[i+1][j]) == "C" {
+				counterGC++
+			}
+			if string(sequenceSpec[i][j]) == "G" && string(sequenceSpec[i+1][j]) == "A" {
+				counterGA++
+			}
+			if string(sequenceSpec[i][j]) == "G" && string(sequenceSpec[i+1][j]) == "T" {
+				counterGT++
+			}
+		}
+	}
+
+	counterCA := 0
+	counterCT := 0
+	counterCG := 0
+
+	for i := 0; i < len(sequenceSpec)-1; i++ {
+		for j := 0; j < len(sequenceSpec[0]); j++ {
+			if string(sequenceSpec[i][j]) == "C" && string(sequenceSpec[i+1][j]) == "A" {
+				counterCA++
+			}
+			if string(sequenceSpec[i][j]) == "C" && string(sequenceSpec[i+1][j]) == "T" {
+				counterCT++
+			}
+			if string(sequenceSpec[i][j]) == "C" && string(sequenceSpec[i+1][j]) == "G" {
+				counterCG++
+			}
+		}
+	}
+
+	fmt.Printf(
+		"The collinearity block for A as a base pattern and T as a mismatch is %d\n",
+		counterAT,
+	)
+	fmt.Printf("The collinearity block for A as a base pattern G as a mismatch is  %d\n", counterAG)
+	fmt.Printf(
+		"The collinearity block for A as a base pattern and C as a mismatch is %d\n",
+		counterAC,
+	)
+
+	fmt.Printf(
+		"The collinearity block for T as a base pattern and G as a mismatch is %d\n",
+		counterTG,
+	)
+	fmt.Printf("The collinearity block for T as a base pattern C as a mismatch is  %d\n", counterTC)
+	fmt.Printf(
+		"The collinearity block for T as a base pattern and A as a mismatch is %d\n",
+		counterTA,
+	)
+
+	fmt.Printf(
+		"The collinearity block for G as a base pattern and C as a mismatch is %d\n",
+		counterGC,
+	)
+	fmt.Printf("The collinearity block for G as a base pattern A as a mismatch is  %d\n", counterGA)
+	fmt.Printf(
+		"The collinearity block for G as a base pattern and T as a mismatch is %d\n",
+		counterGT,
+	)
+
+	fmt.Printf(
+		"The collinearity block for C as a base pattern and A as a mismatch is %d\n",
+		counterCA,
+	)
+	fmt.Printf("The collinearity block for C as a base pattern T as a mismatch is  %d\n", counterCT)
+	fmt.Printf(
+		"The collinearity block for C as a base pattern and G as a mismatch is %d\n",
+		counterCG,
+	)
 }

samplefile/samplealignment.fasta

@@ -1,4 +1,4 @@
 >ENA|OX291461|OX291461.1
-TCAGTATC----TC--------
+TAAGTATC----TC----ATA-
 >ENA|OX291509|OX291509.1
-TATC----TC--------ATAG
+TCTC----TC--------GTAG