(String seq, int windowSize)
| 771 | } |
| 772 | |
| 773 | public static ArrayList<String> trimLowComplexityRegions(String seq, int windowSize) { |
| 774 | ArrayList<String> segments = new ArrayList<>(); |
| 775 | |
| 776 | int seqLen = seq.length(); |
| 777 | |
| 778 | if (seqLen >= windowSize) { |
| 779 | int numWindows = seqLen/windowSize; |
| 780 | boolean[] complexityStatus = new boolean[numWindows]; |
| 781 | |
| 782 | int offset = (seqLen % windowSize) / 2; |
| 783 | int numLowComplexityWindows = 0; |
| 784 | |
| 785 | for (int i=0; i<numWindows; ++i) { |
| 786 | int start = i*windowSize + offset; |
| 787 | int end = start + windowSize; |
| 788 | String window = seq.substring(start, end); |
| 789 | if (isLowComplexityLong(window)) { |
| 790 | ++numLowComplexityWindows; |
| 791 | complexityStatus[i] = false; |
| 792 | } |
| 793 | else { |
| 794 | complexityStatus[i] = true; |
| 795 | } |
| 796 | } |
| 797 | |
| 798 | if (numLowComplexityWindows >= 1) { |
| 799 | int start = -1; |
| 800 | int end = -1; |
| 801 | for (int i=0; i<numWindows; ++i) { |
| 802 | if (complexityStatus[i]) { |
| 803 | if (start < 0) { |
| 804 | start = i; |
| 805 | } |
| 806 | end = i; |
| 807 | } |
| 808 | else if (start >= 0) { |
| 809 | if (start == 0) { |
| 810 | segments.add(seq.substring(0, offset + (end+1) * windowSize)); |
| 811 | } |
| 812 | else { |
| 813 | segments.add(seq.substring(offset + start * windowSize, offset + (end+1) * windowSize)); |
| 814 | } |
| 815 | |
| 816 | start = -1; |
| 817 | end = -1; |
| 818 | } |
| 819 | } |
| 820 | |
| 821 | if (start == 0) { |
| 822 | segments.add(seq.substring(0, seqLen)); |
| 823 | } |
| 824 | else if (start > 0) { |
| 825 | segments.add(seq.substring(offset + start * windowSize, seqLen)); |
| 826 | } |
| 827 | } |
| 828 | else { |
| 829 | segments.add(seq); |
| 830 | } |
no test coverage detected