|
0
|
1 package pairedendCrossPlot;
|
|
|
2
|
|
|
3 import java.io.File;
|
|
|
4 import java.io.FileNotFoundException;
|
|
|
5 import java.io.PrintStream;
|
|
|
6 import java.sql.Timestamp;
|
|
|
7 import java.util.ArrayList;
|
|
|
8 import java.util.Date;
|
|
|
9 import java.util.Scanner;
|
|
|
10
|
|
|
11 public class pairedendCrossPlot {
|
|
|
12
|
|
|
13 private static File READ1 = null;
|
|
|
14 private static File READ2 = null;
|
|
|
15 private static File OUTPUT = null;
|
|
|
16 private static int WINDOW = 500;
|
|
|
17
|
|
|
18 private static int[] BP = null;
|
|
|
19 private static int[] FOR = null;
|
|
|
20 private static int[] REV = null;
|
|
|
21
|
|
|
22 private static ArrayList<String> chrName = null;
|
|
|
23 private static ArrayList<Integer> chrStart = null;
|
|
|
24
|
|
|
25 private static int[] FHIST = null;
|
|
|
26 private static int[] RHIST = null;
|
|
|
27
|
|
|
28 public static void main(String[] args) throws FileNotFoundException {
|
|
|
29 loadConfig(args); //Load configuration data
|
|
|
30 FHIST = new int[(WINDOW * 2) + 1]; //Initialize master histograms
|
|
|
31 RHIST = new int[(WINDOW * 2) + 1]; //Initialize master histograms
|
|
|
32 System.out.println("\n" + getTimeStamp());
|
|
|
33
|
|
|
34 //Pass through the Read2 file once in order to identify the start and stops of all chromosomes present in file
|
|
|
35 System.out.println("Indexing Read 2 File...");
|
|
|
36 indexFile(READ2);
|
|
|
37 System.out.println("Indexing Complete\n" + getTimeStamp());
|
|
|
38
|
|
|
39 //Parse read 1
|
|
|
40 System.out.println("\n" + getTimeStamp() + "\nParsing Read 1 File...");
|
|
|
41 parseRead1(READ1);
|
|
|
42 System.out.println("\nRead 1 Parsed\n" + getTimeStamp());
|
|
|
43
|
|
|
44 //Output histogram
|
|
|
45 System.out.println("\nOutputing final histogram...");
|
|
|
46 outputHist(OUTPUT);
|
|
|
47 System.out.println("Program Complete\n" + getTimeStamp());
|
|
|
48
|
|
|
49 }
|
|
|
50
|
|
|
51 public static void outputHist(File out) throws FileNotFoundException {
|
|
|
52 PrintStream OUT = new PrintStream(out);
|
|
|
53 for(int x = WINDOW * -1; x <= WINDOW; x++) {
|
|
|
54 OUT.print("\t" + x);
|
|
|
55 }
|
|
|
56 OUT.print("\nForward_Tags");
|
|
|
57 for(int x = 0; x < FHIST.length; x++) {
|
|
|
58 OUT.print("\t" + FHIST[x]);
|
|
|
59 }
|
|
|
60 OUT.print("\nReverse_Tags");
|
|
|
61 for(int x = 0; x < RHIST.length; x++) {
|
|
|
62 OUT.print("\t" + RHIST[x]);
|
|
|
63 }
|
|
|
64 OUT.close();
|
|
|
65 }
|
|
|
66
|
|
|
67 public static void parseRead1(File input) throws FileNotFoundException {
|
|
|
68 Scanner scan = new Scanner(READ1);
|
|
|
69 String currentChrom = "";
|
|
|
70 int currentIndex = 0;
|
|
|
71 while (scan.hasNextLine()) {
|
|
|
72 String temp = scan.nextLine();
|
|
|
73 if(!temp.contains("index") && !temp.contains("#")) {
|
|
|
74 String[] array = temp.split("\t");
|
|
|
75 int POS = Integer.parseInt(array[1]);
|
|
|
76 int F = Integer.parseInt(array[2]);
|
|
|
77 int R = Integer.parseInt(array[3]);
|
|
|
78
|
|
|
79 //Check if newline's chrom is equal to current chrom in memory, reload new chrom if unequal
|
|
|
80 if(!currentChrom.equals(array[0])) {
|
|
|
81 BP = null; //reset BP immediately
|
|
|
82 int newIndex = chrName.indexOf(array[0]);
|
|
|
83 if(newIndex >= 0) {
|
|
|
84 System.out.println("\nLoading: " + chrName.get(newIndex));
|
|
|
85 loadRead2(READ2, chrName.get(newIndex), chrStart.get(newIndex), chrStart.get(newIndex + 1));
|
|
|
86 currentChrom = array[0];
|
|
|
87 currentIndex = 0;
|
|
|
88 System.out.println("Loading Complete\n" + getTimeStamp());
|
|
|
89 }
|
|
|
90 }
|
|
|
91 if(BP != null) {
|
|
|
92 //Update local indexes to only upload data from local tag region defined by window
|
|
|
93 while(POS - WINDOW > BP[currentIndex] && currentIndex < BP.length - 1) { currentIndex++; } //Find furthest upstream index in read2
|
|
|
94 int currentStop = currentIndex;
|
|
|
95 while(POS + WINDOW > BP[currentStop] && currentStop < BP.length - 1) { currentStop++; } //Find furthest downstream index in read2
|
|
|
96
|
|
|
97 //Populate histogram based on read2 tags flanking current position
|
|
|
98 int[] temp_F = new int[(WINDOW * 2) + 1];
|
|
|
99 int[] temp_R = new int[(WINDOW * 2) + 1];
|
|
|
100 for(int x = currentIndex; x <= currentStop; x++) {
|
|
|
101 int index = BP[x] - POS + WINDOW;
|
|
|
102 if(index >= 0 && index < temp_F.length) {
|
|
|
103 temp_F[index] += FOR[x];
|
|
|
104 temp_R[index] += REV[x];
|
|
|
105 }
|
|
|
106 }
|
|
|
107
|
|
|
108 //Populate master histogram based on score
|
|
|
109 for(int x = 0; x < temp_F.length; x++) {
|
|
|
110 FHIST[x] += (temp_F[x] * F);
|
|
|
111 RHIST[x] += (temp_R[x] * F);
|
|
|
112 }
|
|
|
113 //Reverse arrays
|
|
|
114 for(int x = 0; x < FHIST.length / 2; x++) {
|
|
|
115 int tempF = temp_F[x];
|
|
|
116 int tempR = temp_R[x];
|
|
|
117 temp_F[x] = temp_F[temp_F.length - x - 1];
|
|
|
118 temp_F[temp_F.length - x - 1] = tempF;
|
|
|
119 temp_R[x] = temp_R[temp_R.length - x - 1];
|
|
|
120 temp_R[temp_R.length - x - 1] = tempR;
|
|
|
121 }
|
|
|
122 for(int x = 0; x < FHIST.length; x++) {
|
|
|
123 FHIST[x] += (temp_R[x] * R);
|
|
|
124 RHIST[x] += (temp_F[x] * R);
|
|
|
125 }
|
|
|
126 }
|
|
|
127 }
|
|
|
128 }
|
|
|
129 scan.close();
|
|
|
130 }
|
|
|
131
|
|
|
132 private static void loadRead2(File input, String currentChrom, int lineStart, int lineStop) throws FileNotFoundException {
|
|
|
133 BP = new int[lineStop - lineStart - 1];
|
|
|
134 FOR = new int[lineStop - lineStart - 1];
|
|
|
135 REV = new int[lineStop - lineStart - 1];
|
|
|
136
|
|
|
137 int counter = 0;
|
|
|
138 Scanner scan = new Scanner(READ2);
|
|
|
139 while (scan.hasNextLine()) {
|
|
|
140 String temp = scan.nextLine();
|
|
|
141 if(counter > lineStart) {
|
|
|
142 String[] array = temp.split("\t");
|
|
|
143 if(array[0].equals(currentChrom)) {
|
|
|
144 BP[counter - lineStart - 1] = Integer.parseInt(array[1]);
|
|
|
145 FOR[counter - lineStart - 1] = Integer.parseInt(array[2]);
|
|
|
146 REV[counter - lineStart - 1] = Integer.parseInt(array[3]);
|
|
|
147 }
|
|
|
148 }
|
|
|
149 counter++;
|
|
|
150 }
|
|
|
151 scan.close();
|
|
|
152 //System.out.println(BP.length + "\t" + FOR.length + "\t" + REV.length);
|
|
|
153 }
|
|
|
154
|
|
|
155 //Index Read 2 file for faster processing of chromosomes
|
|
|
156 public static void indexFile(File input) throws FileNotFoundException {
|
|
|
157 chrName = new ArrayList<String>();
|
|
|
158 chrStart = new ArrayList<Integer>();
|
|
|
159 Scanner scan = new Scanner(input);
|
|
|
160 int linenumber = 0;
|
|
|
161 while (scan.hasNextLine()) {
|
|
|
162 String temp = scan.nextLine();
|
|
|
163 if(!temp.contains("index") && !temp.contains("#")) {
|
|
|
164 String[] array = temp.split("\t");
|
|
|
165 if(!chrName.contains(array[0])) {
|
|
|
166 chrName.add(array[0]);
|
|
|
167 chrStart.add(new Integer(linenumber));
|
|
|
168 }
|
|
|
169 }
|
|
|
170 linenumber++;
|
|
|
171 }
|
|
|
172 chrStart.add(new Integer(linenumber));
|
|
|
173 scan.close();
|
|
|
174 }
|
|
|
175
|
|
|
176 public static void loadConfig(String[] command){
|
|
|
177 for (int i = 0; i < command.length; i++) {
|
|
|
178 switch (command[i].charAt((1))) {
|
|
|
179 case 'f':
|
|
|
180 READ1 = new File(command[i + 1]);
|
|
|
181 i++;
|
|
|
182 break;
|
|
|
183 case 'r':
|
|
|
184 READ2 = new File(command[i + 1]);
|
|
|
185 i++;
|
|
|
186 break;
|
|
|
187 case 'o':
|
|
|
188 OUTPUT = new File(command[i + 1]);
|
|
|
189 i++;
|
|
|
190 break;
|
|
|
191 case 'w':
|
|
|
192 WINDOW = Integer.parseInt(command[i + 1]);
|
|
|
193 i++;
|
|
|
194 break;
|
|
|
195 }
|
|
|
196 }
|
|
|
197 if(READ1 == null) {
|
|
|
198 System.out.println("No Read 1 File loaded!!!\n");
|
|
|
199 printUsage();
|
|
|
200 System.exit(0);
|
|
|
201 }
|
|
|
202 if(READ2 == null) {
|
|
|
203 System.out.println("No Read 2 File loaded!!!\n");
|
|
|
204 printUsage();
|
|
|
205 System.exit(0);
|
|
|
206 }
|
|
|
207 if(WINDOW < 1) {
|
|
|
208 System.out.println("Invalid Window Size Selected!!!\n");
|
|
|
209 printUsage();
|
|
|
210 System.exit(0);
|
|
|
211 }
|
|
|
212
|
|
|
213 if(OUTPUT == null) {
|
|
|
214 OUTPUT = new File(READ1.getName().split("\\.")[0] + ".out");
|
|
|
215 }
|
|
|
216
|
|
|
217 System.out.println("-----------------------------------------\nCommand Line Arguments:");
|
|
|
218 System.out.println("Read 1 file: " + READ1);
|
|
|
219 System.out.println("Read 2 file: " + READ2);
|
|
|
220 System.out.println("Output: " + OUTPUT);
|
|
|
221 System.out.println("Window Size: " + WINDOW);
|
|
|
222 }
|
|
|
223
|
|
|
224 public static void printUsage() {
|
|
|
225 System.out.println("Usage: java -jar pairedendCrossPlot.jar -f [Read1File] -r [Read2File] [Options]");
|
|
|
226 System.out.println("-----------------------------------------");
|
|
|
227 System.out.println("\nRequired Parameter:");
|
|
|
228 System.out.println("Read 1 File:\t\t\t-f\tRead 1 scIDX file");
|
|
|
229 System.out.println("Read 2 File:\t\t\t-r\tRead 1 scIDX file");
|
|
|
230 System.out.println("\nOptional Parameters:");
|
|
|
231 System.out.println("Output File Name:\t\t-o\tOutput file");
|
|
|
232 System.out.println("Window Size +/- Reference:\t-w\t(Default = 500)");
|
|
|
233 }
|
|
|
234
|
|
|
235 private static String getTimeStamp() {
|
|
|
236 Date date= new Date();
|
|
|
237 String time = new Timestamp(date.getTime()).toString();
|
|
|
238 return time;
|
|
|
239 }
|
|
|
240
|
|
|
241 }
|
