-
-
Save elvismdev/12ba4e6efc01730e193c to your computer and use it in GitHub Desktop.
| /** | |
| * Find all the genes from a DNA string file and using StorageResource class. | |
| * | |
| * @author (Elvis Morales) | |
| * @version (1.0) | |
| */ | |
| import edu.duke.*; | |
| import java.io.File; | |
| public class FindMultipleGenesStorage { | |
| public int findStopIndex(String dna, int index) { | |
| int stop1 = dna.indexOf("tga", index); | |
| if ( stop1 == -1 || ( stop1-index ) % 3 != 0 ) { | |
| stop1 = dna.length(); | |
| } | |
| int stop2 = dna.indexOf("taa", index); | |
| if ( stop2 == -1 || ( stop2-index ) % 3 != 0 ) { | |
| stop2 = dna.length(); | |
| } | |
| int stop3 = dna.indexOf("tag", index); | |
| if ( stop3 == -1 || ( stop3-index ) % 3 != 0 ) { | |
| stop3 = dna.length(); | |
| } | |
| return Math.min( stop1, Math.min(stop2, stop3) ); | |
| } | |
| public StorageResource storeAll(String dna) { | |
| String dnaLow = dna.toLowerCase(); | |
| int start = 0; | |
| StorageResource genes = new StorageResource(); | |
| while (true) { | |
| int loc = dnaLow.indexOf( "atg", start ); | |
| if ( loc == -1 ) { | |
| break; | |
| } | |
| int stop = findStopIndex( dnaLow, loc+3 ); | |
| if ( stop != dna.length() ) { | |
| genes.add( dna.substring(loc, stop+3) ); | |
| start = stop + 3; | |
| } else { | |
| start = start + 3; | |
| } | |
| } | |
| return genes; | |
| } | |
| public void testStorageFinder() { | |
| FileResource dnaFile = new FileResource(); | |
| StorageResource genesFound = storeAll( dnaFile.asString() ); | |
| System.out.println( "Number of genes found: "+genesFound.size() ); | |
| printGenes( genesFound ); | |
| } | |
| public float cgRatio( String dna ) { | |
| String dnaLow = dna.toLowerCase(); | |
| int cgCount = 0; | |
| int start = 0; | |
| while (true) { | |
| int pos = dnaLow.indexOf("c", start); | |
| if (pos == -1) { | |
| start = 0; | |
| break; | |
| } | |
| cgCount += 1; | |
| start = pos + 1; | |
| } | |
| while (true) { | |
| int pos = dnaLow.indexOf("g", start); | |
| if (pos == -1) { | |
| start = 0; | |
| break; | |
| } | |
| cgCount += 1; | |
| start = pos + 1; | |
| } | |
| return ( (float) cgCount ) / dna.length(); | |
| } | |
| public void printGenes( StorageResource sr ) { | |
| int sixtyCharQty = 0; | |
| int highCgRatioQty = 0; | |
| float cgRatioConst = (float) 0.35; | |
| for ( String s : sr.data() ) { | |
| if ( s.length() > 60 ) { | |
| System.out.println( "String longer than 60 characters: "+s ); | |
| sixtyCharQty++; | |
| } | |
| if ( cgRatio(s) > cgRatioConst ) { | |
| System.out.println( "String with C-G-ratio higher than 0.35: "+s ); | |
| highCgRatioQty++; | |
| } | |
| } | |
| System.out.println( "60 characters qty: "+sixtyCharQty ); | |
| System.out.println( "Strings with C-G-ratio higher than 0.35: "+highCgRatioQty ); | |
| } | |
| } |
/**
import edu.duke.*;
public class part1 {
public int findStopCodon(String dna,int startIndex,String stopcodon ){
int currIndex = dna.indexOf(stopcodon,startIndex+3);
while(currIndex!=-1){
if((currIndex-startIndex)%3==0){
return currIndex;
}
else{
currIndex = dna.indexOf(stopcodon,currIndex+1);
}
}
return dna.length();
}
public String findGene(String dna ){
int startIndex = dna.indexOf("ATG");
if(startIndex==-1){
return "" ;
}
int indexTAA = findStopCodon(dna,startIndex,"TAA");
int indexTAG = findStopCodon(dna,startIndex,"TAG");
int indexTGA = findStopCodon(dna,startIndex,"TGA");
int temp= Math.min(indexTAA,indexTAG);
int firstStopCodon = Math.min(temp,indexTGA);
if(firstStopCodon==dna.length()){
return "";
}
String resultString = dna.substring(startIndex,firstStopCodon+3);//line 40
return resultString;
}
public StorageResource getAllGene(String dna){
StorageResource geneList = new StorageResource();
int startIndex = 0;
while(true){
String currentGene = findGene(dna);//line 50
if(currentGene.length()==0){
break;
}else{
geneList.add(currentGene);
startIndex = dna.indexOf(currentGene,startIndex)+currentGene.length();
dna = dna.substring(startIndex,dna.length());
}
}
return geneList;
}
public double cgRatio(String dna){
StorageResource gene = getAllGene(dna);
double cgratio = 0;
for(String g: gene.data()){
int cOcc=0,oOcc=0,startIndexC=0,startIndexO=0;
double count=0;
while(cOcc!=-1||oOcc!=-1){
cOcc=dna.indexOf("C",startIndexC);
if(cOcc!=-1){
count++;
startIndexC=cOcc+1;
}
oOcc=dna.indexOf("G",startIndexO);
if(oOcc!=-1){
count++;
startIndexO=oOcc+1;
}
}
double dnaLength = dna.length();
cgratio = count/dnaLength;
return cgratio;
}
return cgratio;
}
public int countCTG(String dna){
int count =0, ctgOcc =0,startIndexCTG=0;
while (ctgOcc!=-1){
ctgOcc=dna.indexOf("CTG",startIndexCTG);
if(ctgOcc!=-1){
count++;
startIndexCTG=ctgOcc+3;
}
}
return count;
}
public void processGenes(StorageResource sr){
int count = 0,countCg = 0,countgene=1;
int temp=0;
for( String g: sr.data()){
StorageResource geneList=getAllGene(g); //line 116
for(String gList:geneList.data()){
System.out.println("gene "+countgene +" cointained is "+gList);
if(gList.length()>60){
System.out.println("The gene with a length longer then 60 is "+gList);
count++;
}
if(temp<gList.length()){
temp=gList.length();
}
else{
temp=temp;
}
countgene++;
double cgRatio=cgRatio(gList);
if( cgRatio>0.35)
{
countCg++;
System.out.println("the "+countCg+" cg ratio is "+cgRatio);
}
}
int numberOfCTG=countCTG(g);
System.out.println("the number of time CTG appears is "+numberOfCTG);
}
System.out.println("the number of string with length longer then 60 is "+count);
System.out.println("the no of string with cgratio greater then 0.35 "+countCg);
System.out.println("the longest length of gene in the dna is "+temp);
}
public void testProcessGene(){
FileResource fr=new FileResource("reviewTest.fa");
String dna1 = fr.asString();
String dna = dna1.toUpperCase();
System.out.println(dna);
StorageResource sr= new StorageResource();
sr.add(dna);
processGenes(sr);//line 152
}
}
// reviewTest.fa=https://users.cs.duke.edu/~rodger/GRch38dnapart.fa
//please help me why my answere is not correct where is the bug in my code
// I am getting total no of gene 34 and the answere is 69 help me
//Total number of strings with length greater than 60 : 23(I am getting 11)
//Total number of strings with CG Ratio greater than 0.35 : 40(i am getting 25)
//Longest gene length is : 489 (i got this right)
//Total number of occurances of CTG is : 224 (i got this right)
please help me
TOTAL GENES = 69
Total number of strings with length greater than 60 : 23
Total number of strings with CG Ratio greater than 0.35 : 40
Longest gene length is : 489
Total number of occurances of CTG is : 224Cheers!
Thanks! All correct. Actually I was stuck at the longest gene length.
can i get week 3 practice quiz and final quiz answer ?
TOTAL GENES = 69
Total number of strings with length greater than 60 : 23
Total number of strings with CG Ratio greater than 0.35 : 40
Longest gene length is : 489
Total number of occurances of CTG is : 224Cheers!
Can I get your code bro?
Week 3 assignment answers please
How many genes are there in the file brca1line.fa?
Ans 1
How many genes are there in the file brca1line.fa that are longer than 60?
Ans 1
How many genes are there in the file brca1line.fa that have a C-G-ratio greater than 0.35?
Ans 1
How many genes are there in the file brca1line.fa?
Ans 1
How many genes are there in the file brca1line.fa that are longer than 60?
Ans 1
How many genes are there in the file brca1line.fa that have a C-G-ratio greater than 0.35?
Ans 1
Right one
TOTAL GENES = 69
Total number of strings with length greater than 60 : 23
Total number of strings with CG Ratio greater than 0.35 : 40
Longest gene length is : 489
Total number of occurances of CTG is : 224Cheers!
Hey, can you please send the code...
TOTAL GENES = 69
Total number of strings with length greater than 60 : 23
Total number of strings with CG Ratio greater than 0.35 : 40
Longest gene length is : 489
Total number of occurances of CTG is : 224Cheers!
Wrong answers
I don't understand why the answers are all ones, I'm finding some valid genes with my code and having them verified with my bare eyes.
Aapne bachaliya bhai