toRange, int fromRatio,
+ int toRatio)
+ {
+ this();
+ fromRange = coalesceRanges(fromRange);
+ toRange = coalesceRanges(toRange);
+ this.fromShifts = fromRange;
+ this.toShifts = toRange;
+ this.fromRatio = fromRatio;
+ this.toRatio = toRatio;
+
+ fromLowest = Integer.MAX_VALUE;
+ fromHighest = Integer.MIN_VALUE;
+ for (int[] range : fromRange)
+ {
+ if (range.length != 2)
+ {
+ // throw new IllegalArgumentException(range);
+ Cache.log.error("Invalid format for fromRange "
+ + Arrays.toString(range) + " may cause errors");
+ }
+ fromLowest = Math.min(fromLowest, Math.min(range[0], range[1]));
+ fromHighest = Math.max(fromHighest, Math.max(range[0], range[1]));
+ }
+
+ toLowest = Integer.MAX_VALUE;
+ toHighest = Integer.MIN_VALUE;
+ for (int[] range : toRange)
+ {
+ if (range.length != 2)
+ {
+ // throw new IllegalArgumentException(range);
+ Cache.log.error("Invalid format for toRange "
+ + Arrays.toString(range) + " may cause errors");
+ }
+ toLowest = Math.min(toLowest, Math.min(range[0], range[1]));
+ toHighest = Math.max(toHighest, Math.max(range[0], range[1]));
+ }
+ }
+
+ /**
+ * Consolidates a list of ranges so that any contiguous ranges are merged.
+ * This assumes the ranges are already in start order (does not sort them).
+ *
+ * The main use case for this method is when mapping cDNA sequence to its
+ * protein product, based on CDS feature ranges which derive from spliced
+ * exons, but are contiguous on the cDNA sequence. For example
+ *
+ * CDS 1-20 // from exon1
+ * CDS 21-35 // from exon2
+ * CDS 36-71 // from exon3
+ * 'coalesce' to range 1-71
+ *
+ *
+ * @param ranges
+ * @return the same list (if unchanged), else a new merged list, leaving the
+ * input list unchanged
+ */
+ public static List coalesceRanges(final List ranges)
+ {
+ if (ranges == null || ranges.size() < 2)
+ {
+ return ranges;
+ }
+
+ boolean changed = false;
+ List merged = new ArrayList<>();
+ int[] lastRange = ranges.get(0);
+ int lastDirection = lastRange[1] >= lastRange[0] ? 1 : -1;
+ lastRange = new int[] { lastRange[0], lastRange[1] };
+ merged.add(lastRange);
+ boolean first = true;
+
+ for (final int[] range : ranges)
+ {
+ if (first)
+ {
+ first = false;
+ continue;
+ }
+
+ int direction = range[1] >= range[0] ? 1 : -1;
+
+ /*
+ * if next range is in the same direction as last and contiguous,
+ * just update the end position of the last range
+ */
+ boolean sameDirection = range[1] == range[0]
+ || direction == lastDirection;
+ boolean extending = range[0] == lastRange[1] + lastDirection;
+ if (sameDirection && extending)
+ {
+ lastRange[1] = range[1];
+ changed = true;
+ }
+ else
+ {
+ lastRange = new int[] { range[0], range[1] };
+ merged.add(lastRange);
+ // careful: merging [5, 5] after [7, 6] should keep negative direction
+ lastDirection = (range[1] == range[0]) ? lastDirection : direction;
+ }
+ }
+
+ return changed ? merged : ranges;
+ }
+
/**
* get all mapped positions from 'from' to 'to'
- * @return int[][] { int[] { fromStart, fromFinish, toStart, toFinish }, int [fromFinish-fromStart+2] { toStart..toFinish mappings}}
+ *
+ * @return int[][] { int[] { fromStart, fromFinish, toStart, toFinish }, int
+ * [fromFinish-fromStart+2] { toStart..toFinish mappings}}
*/
- public int[][] makeFromMap()
+ protected int[][] makeFromMap()
{
+ // TODO not used - remove??
return posMap(fromShifts, fromRatio, toShifts, toRatio);
}
+
/**
* get all mapped positions from 'to' to 'from'
+ *
* @return int[to position]=position mapped in from
*/
- public int[][] makeToMap()
+ protected int[][] makeToMap()
{
- return posMap(toShifts,toRatio, fromShifts, fromRatio);
+ // TODO not used - remove??
+ return posMap(toShifts, toRatio, fromShifts, fromRatio);
}
+
/**
* construct an int map for intervals in intVals
- * @param intVals
- * @return int[] { from, to pos in range }, int[range.to-range.from+1] returning mapped position
+ *
+ * @param shiftTo
+ * @return int[] { from, to pos in range }, int[range.to-range.from+1]
+ * returning mapped position
*/
- private int[][] posMap(Vector intVals, int ratio, Vector toIntVals,
- int toRatio)
+ private int[][] posMap(List shiftTo, int ratio,
+ List shiftFrom, int toRatio)
{
- int iv=0,ivSize = intVals.size();
- if (iv>=ivSize)
+ // TODO not used - remove??
+ int iv = 0, ivSize = shiftTo.size();
+ if (iv >= ivSize)
{
return null;
}
- int[] intv=(int[]) intVals.elementAt(iv++);
- int from=intv[0],to=intv[1];
+ int[] intv = shiftTo.get(iv++);
+ int from = intv[0], to = intv[1];
if (from > to)
{
from = intv[1];
- to=intv[0];
+ to = intv[0];
}
- while (ivto)
+ if (intv[0] > to)
{
- to=intv[0];
+ to = intv[0];
}
- if (intv[1]>to)
+ if (intv[1] > to)
{
- to=intv[1];
+ to = intv[1];
}
}
- int tF=0,tT=0;
- int mp[][] = new int[to-from+2][];
+ int tF = 0, tT = 0;
+ int mp[][] = new int[to - from + 2][];
for (int i = 0; i < mp.length; i++)
{
- int[] m = shift(i+from,intVals,ratio,toIntVals, toRatio);
+ int[] m = shift(i + from, shiftTo, ratio, shiftFrom, toRatio);
if (m != null)
{
if (i == 0)
{
- tF=tT=m[0];
+ tF = tT = m[0];
}
else
{
if (m[0] < tF)
{
- tF=m[0];
+ tF = m[0];
}
if (m[0] > tT)
{
- tT=m[0];
+ tT = m[0];
}
}
}
mp[i] = m;
}
- int[][] map = new int[][]
- {
- new int[]
- {
- from, to, tF, tT}, new int[to - from + 2]};
+ int[][] map = new int[][] { new int[] { from, to, tF, tT },
+ new int[to - from + 2] };
map[0][2] = tF;
map[0][3] = tT;
@@ -284,7 +500,7 @@ public class MapList
{
if (mp[i] != null)
{
- map[1][i] = mp[i][0]-tF;
+ map[1][i] = mp[i][0] - tF;
}
else
{
@@ -293,28 +509,29 @@ public class MapList
}
return map;
}
+
/**
* addShift
- * @param pos start position for shift (in original reference frame)
- * @param shift length of shift
- *
- public void addShift(int pos, int shift)
- {
- int sidx = 0;
- int[] rshift=null;
- while (sidx shiftTo, int fromRatio,
+ List shiftFrom, int toRatio)
{
- int[] fromCount = countPos(fromShifts, pos);
- if (fromCount==null)
+ // TODO: javadoc; tests
+ int[] fromCount = countPos(shiftTo, pos);
+ if (fromCount == null)
{
return null;
}
- int fromRemainder=(fromCount[0]-1) % fromRatio;
- int toCount = 1+(((fromCount[0]-1) / fromRatio) * toRatio);
- int[] toPos = countToPos(toShifts, toCount);
- if (toPos==null)
+ int fromRemainder = (fromCount[0] - 1) % fromRatio;
+ int toCount = 1 + (((fromCount[0] - 1) / fromRatio) * toRatio);
+ int[] toPos = countToPos(shiftFrom, toCount);
+ if (toPos == null)
{
return null; // throw new Error("Bad Mapping!");
}
- //System.out.println(fromCount[0]+" "+fromCount[1]+" "+toCount);
- return new int[]
- {
- toPos[0], fromRemainder, toPos[1]};
+ // System.out.println(fromCount[0]+" "+fromCount[1]+" "+toCount);
+ return new int[] { toPos[0], fromRemainder, toPos[1] };
}
+
/**
* count how many positions pos is along the series of intervals.
- * @param intVals
+ *
+ * @param shiftTo
* @param pos
* @return number of positions or null if pos is not within intervals
*/
- private int[] countPos(Vector intVals, int pos)
+ protected static int[] countPos(List shiftTo, int pos)
{
- int count=0,intv[],iv=0,ivSize=intVals.size();
- while (iv= intv[0] && pos <= intv[1])
{
- return new int[]
- {
- count + pos - intv[0] + 1, +1};
+ return new int[] { count + pos - intv[0] + 1, +1 };
}
else
{
- count+=intv[1]-intv[0]+1;
+ count += intv[1] - intv[0] + 1;
}
}
else
{
if (pos >= intv[1] && pos <= intv[0])
{
- return new int[]
- {
- count + intv[0] - pos + 1, -1};
+ return new int[] { count + intv[0] - pos + 1, -1 };
}
else
{
- count+=intv[0]-intv[1]+1;
+ count += intv[0] - intv[1] + 1;
}
}
}
return null;
}
+
/**
* count out pos positions into a series of intervals and return the position
- * @param intVals
+ *
+ * @param shiftFrom
* @param pos
* @return position pos in interval set
*/
- private int[] countToPos(Vector intVals, int pos)
+ protected static int[] countToPos(List shiftFrom, int pos)
{
- int count = 0, diff = 0, iv=0,ivSize=intVals.size(), intv[] =
- {
- 0, 0};
- while (iv= 0)
{
if (pos <= count + 1 + diff)
{
- return new int[]
- {
- pos - count - 1 + intv[0], +1};
+ return new int[] { pos - count - 1 + intv[0], +1 };
}
else
{
- count+=1+diff;
+ count += 1 + diff;
}
}
else
{
if (pos <= count + 1 - diff)
{
- return new int[]
- {
- intv[0] - (pos - count - 1), -1};
+ return new int[] { intv[0] - (pos - count - 1), -1 };
}
else
{
- count+=1-diff;
+ count += 1 - diff;
}
}
}
- return null;//(diff<0) ? (intv[1]-1) : (intv[0]+1);
+ return null;// (diff<0) ? (intv[1]-1) : (intv[0]+1);
}
+
/**
* find series of intervals mapping from start-end in the From map.
- * @param start position in to map
- * @param end position in to map
- * @return series of ranges in from map
+ *
+ * @param start
+ * position mapped 'to'
+ * @param end
+ * position mapped 'to'
+ * @return series of [start, end] ranges in sequence mapped 'from'
*/
- public int[] locateInFrom(int start, int end) {
+ public int[] locateInFrom(int start, int end)
+ {
// inefficient implementation
int fromStart[] = shiftTo(start);
- int fromEnd[] = shiftTo(end); // needs to be inclusive of end of symbol position
- if (fromStart==null || fromEnd==null)
- return null;
- int iv[] = getIntervals(fromShifts, fromStart, fromEnd,fromRatio);
- return iv;
+ // needs to be inclusive of end of symbol position
+ int fromEnd[] = shiftTo(end);
+
+ return getIntervals(fromShifts, fromStart, fromEnd, fromRatio);
}
/**
* find series of intervals mapping from start-end in the to map.
- * @param start position in from map
- * @param end position in from map
- * @return series of ranges in to map
+ *
+ * @param start
+ * position mapped 'from'
+ * @param end
+ * position mapped 'from'
+ * @return series of [start, end] ranges in sequence mapped 'to'
*/
- public int[] locateInTo(int start, int end) {
- // inefficient implementation
+ public int[] locateInTo(int start, int end)
+ {
int toStart[] = shiftFrom(start);
int toEnd[] = shiftFrom(end);
- if (toStart==null || toEnd==null)
- return null;
- int iv[] = getIntervals(toShifts, toStart, toEnd, toRatio);
- return iv;
+ return getIntervals(toShifts, toStart, toEnd, toRatio);
}
+
/**
- * like shift - except returns the intervals in the given vector of shifts which were spanned
- * in traversing fromStart to fromEnd
- * @param fromShifts2
+ * like shift - except returns the intervals in the given vector of shifts
+ * which were spanned in traversing fromStart to fromEnd
+ *
+ * @param shiftFrom
* @param fromStart
* @param fromEnd
* @param fromRatio2
- * @return series of from,to intervals from from first position of starting region to final position of ending region inclusive
- */
- private int[] getIntervals(Vector fromShifts2, int[] fromStart, int[] fromEnd, int fromRatio2)
- {
- int startpos,endpos;
- startpos = fromStart[0]; // first position in fromStart
- endpos = fromEnd[0]+fromEnd[2]*(fromRatio2-1); // last position in fromEnd
- int intv=0,intvSize= fromShifts2.size();
- int iv[],i=0,fs=-1,fe=-1; // containing intervals
- while (intv=iv[0] && startpos<=iv[1]) {
+ * @return series of from,to intervals from from first position of starting
+ * region to final position of ending region inclusive
+ */
+ protected static int[] getIntervals(List shiftFrom,
+ int[] fromStart, int[] fromEnd, int fromRatio2)
+ {
+ if (fromStart == null || fromEnd == null)
+ {
+ return null;
+ }
+ int startpos, endpos;
+ startpos = fromStart[0]; // first position in fromStart
+ endpos = fromEnd[0]; // last position in fromEnd
+ int endindx = (fromRatio2 - 1); // additional positions to get to last
+ // position from endpos
+ int intv = 0, intvSize = shiftFrom.size();
+ int iv[], i = 0, fs = -1, fe_s = -1, fe = -1; // containing intervals
+ // search intervals to locate ones containing startpos and count endindx
+ // positions on from endpos
+ while (intv < intvSize && (fs == -1 || fe == -1))
+ {
+ iv = shiftFrom.get(intv++);
+ if (fe_s > -1)
+ {
+ endpos = iv[0]; // start counting from beginning of interval
+ endindx--; // inclusive of endpos
+ }
+ if (iv[0] <= iv[1])
+ {
+ if (fs == -1 && startpos >= iv[0] && startpos <= iv[1])
+ {
fs = i;
}
- if (fe==-1 && endpos>=iv[0] && endpos<=iv[1]) {
- fe = i;
+ if (endpos >= iv[0] && endpos <= iv[1])
+ {
+ if (fe_s == -1)
+ {
+ fe_s = i;
+ }
+ if (fe_s != -1)
+ {
+ if (endpos + endindx <= iv[1])
+ {
+ fe = i;
+ endpos = endpos + endindx; // end of end token is within this
+ // interval
+ }
+ else
+ {
+ endindx -= iv[1] - endpos; // skip all this interval too
+ }
+ }
}
- } else {
- if (fs==-1 && startpos<=iv[0] && startpos>=iv[1]) {
+ }
+ else
+ {
+ if (fs == -1 && startpos <= iv[0] && startpos >= iv[1])
+ {
fs = i;
}
- if (fe==-1 && endpos<=iv[0] && endpos>=iv[1]) {
- fe = i;
+ if (endpos <= iv[0] && endpos >= iv[1])
+ {
+ if (fe_s == -1)
+ {
+ fe_s = i;
+ }
+ if (fe_s != -1)
+ {
+ if (endpos - endindx >= iv[1])
+ {
+ fe = i;
+ endpos = endpos - endindx; // end of end token is within this
+ // interval
+ }
+ else
+ {
+ endindx -= endpos - iv[1]; // skip all this interval too
+ }
+ }
}
}
i++;
}
- if (fs==fe && fe==-1)
+ if (fs == fe && fe == -1)
+ {
return null;
- Vector ranges=new Vector();
- if (fs<=fe) {
+ }
+ List ranges = new ArrayList<>();
+ if (fs <= fe)
+ {
intv = fs;
- i=fs;
+ i = fs;
// truncate initial interval
- iv = (int[]) fromShifts2.elementAt(intv++);
- iv = new int[] { iv[0], iv[1]};// clone
- if (i==fs)
+ iv = shiftFrom.get(intv++);
+ iv = new int[] { iv[0], iv[1] };// clone
+ if (i == fs)
+ {
iv[0] = startpos;
- while (i!=fe) {
- ranges.addElement(iv); // add initial range
- iv = (int[]) fromShifts2.elementAt(intv++); // get next interval
- iv = new int[] { iv[0], iv[1]};// clone
+ }
+ while (i != fe)
+ {
+ ranges.add(iv); // add initial range
+ iv = shiftFrom.get(intv++); // get next interval
+ iv = new int[] { iv[0], iv[1] };// clone
i++;
}
- if (i==fe)
+ if (i == fe)
+ {
iv[1] = endpos;
- ranges.addElement(iv); // add only - or final range
- } else {
+ }
+ ranges.add(iv); // add only - or final range
+ }
+ else
+ {
// walk from end of interval.
- i=fromShifts2.size()-1;
- while (i>fs) {
+ i = shiftFrom.size() - 1;
+ while (i > fs)
+ {
i--;
}
- iv = (int[]) fromShifts2.elementAt(i);
- iv = new int[] { iv[1], iv[0]};// reverse and clone
+ iv = shiftFrom.get(i);
+ iv = new int[] { iv[1], iv[0] };// reverse and clone
// truncate initial interval
- if (i==fs)
+ if (i == fs)
{
iv[0] = startpos;
}
- while (--i!=fe) { // fix apparent logic bug when fe==-1
- ranges.addElement(iv); // add (truncated) reversed interval
- iv = (int[]) fromShifts2.elementAt(i);
+ while (--i != fe)
+ { // fix apparent logic bug when fe==-1
+ ranges.add(iv); // add (truncated) reversed interval
+ iv = shiftFrom.get(i);
iv = new int[] { iv[1], iv[0] }; // reverse and clone
}
- if (i==fe) {
+ if (i == fe)
+ {
// interval is already reversed
iv[1] = endpos;
}
- ranges.addElement(iv); // add only - or final range
+ ranges.add(iv); // add only - or final range
}
// create array of start end intervals.
int[] range = null;
- if (ranges!=null && ranges.size()>0)
+ if (ranges != null && ranges.size() > 0)
{
- range = new int[ranges.size()*2];
+ range = new int[ranges.size() * 2];
intv = 0;
- intvSize=ranges.size();
- i=0;
- while (intv" + toofrom[0]);
- }
- System.out.println("ShiftTo(" + too[0] + ")==" + toofrom[0] + " % " +
- toofrom[1]+" ("+toofrom[2]+")");
- }
- else
- {
- System.out.println("ShiftTo(" + too[0] + ")==" +
- "NaN! - not Bijective Mapping!");
- }
- }
+ return ((getFromLowest() >= map.getFromLowest()
+ && getFromHighest() <= map.getFromHighest())
+ || (getFromLowest() <= map.getFromLowest()
+ && getFromHighest() >= map.getFromHighest()));
}
- int mmap[][] = ml.makeFromMap();
- System.out.println("FromMap : (" + mmap[0][0] + " " + mmap[0][1] + " " +
- mmap[0][2] + " " + mmap[0][3] + " ");
- for (int i = 1; i <= mmap[1].length; i++)
+ else
{
- if (mmap[1][i - 1] == -1)
- {
- System.out.print(i+"=XXX");
+ return ((getToLowest() >= map.getToLowest()
+ && getToHighest() <= map.getToHighest())
+ || (getToLowest() <= map.getToLowest()
+ && getToHighest() >= map.getToHighest()));
+ }
+ }
- }
- else
- {
- System.out.print(i+"="+(mmap[0][2]+mmap[1][i-1]));
- }
- if (i % 20==0)
- {
- System.out.print("\n");
- }
- else
- {
- System.out.print(",");
- }
+ /**
+ * String representation - for debugging, not guaranteed not to change
+ */
+ @Override
+ public String toString()
+ {
+ StringBuilder sb = new StringBuilder(64);
+ sb.append("[");
+ for (int[] shift : fromShifts)
+ {
+ sb.append(" ").append(Arrays.toString(shift));
}
- //test range function
- System.out.print("\nTest locateInFrom\n");
+ sb.append(" ] ");
+ sb.append(fromRatio).append(":").append(toRatio);
+ sb.append(" to [");
+ for (int[] shift : toShifts)
{
- int f=mmap[0][2],t=mmap[0][3];
- while (f<=t) {
- System.out.println("Range "+f+" to "+t);
- int rng[] = ml.locateInFrom(f,t);
- if (rng!=null)
- {
- for (int i=0; i
+ * Returns null if the mappings cannot be traversed (not all toRanges of this
+ * map correspond to fromRanges of the input), or if this.toRatio does not
+ * match map.fromRatio.
+ *
+ *
+ * Example 1:
+ * this: from [1-100] to [501-600]
+ * input: from [10-40] to [60-90]
+ * output: from [10-40] to [560-590]
+ * Example 2 ('reverse strand exons'):
+ * this: from [1-100] to [2000-1951], [1000-951] // transcript to loci
+ * input: from [1-50] to [41-90] // CDS to transcript
+ * output: from [10-40] to [1960-1951], [1000-971] // CDS to gene loci
+ *
+ *
+ * @param map
+ * @return
+ */
+ public MapList traverse(MapList map)
+ {
+ if (map == null)
{
- int f=mmap[0][2],t=mmap[0][3];
- while (f<=t) {
- System.out.println("Range "+f+" to "+t);
- int rng[] = ml.locateInTo(f,t);
- if (rng!=null) {
- for (int i=0; i toRanges = new ArrayList<>();
+ for (int[] range : getToRanges())
+ {
+ int[] transferred = map.locateInTo(range[0], range[1]);
+ if (transferred == null || transferred.length % 2 != 0)
+ {
+ return null;
+ }
+
+ /*
+ * convert [start1, end1, start2, end2, ...]
+ * to [[start1, end1], [start2, end2], ...]
+ */
+ for (int i = 0; i < transferred.length;)
+ {
+ toRanges.add(new int[] { transferred[i], transferred[i + 1] });
+ i += 2;
}
}
+ return new MapList(getFromRanges(), toRanges, outFromRatio, outToRatio);
}
- public static void main(String argv[])
+ /**
+ * Answers true if the mapping is from one contiguous range to another, else
+ * false
+ *
+ * @return
+ */
+ public boolean isContiguous()
{
- MapList ml = new MapList(new int[]
- {1, 5, 10, 15, 25, 20},
- new int[]
- {51, 1}, 1, 3);
- MapList ml1 = new MapList(new int[]
- {1, 3, 17, 4},
- new int[]
- {51, 1}, 1, 3);
- MapList ml2 = new MapList(new int[] { 1, 60 },
- new int[] { 1, 20 }, 3, 1);
- // test internal consistency
- int to[] = new int[51];
- MapList.testMap(ml, 1, 60);
- /*
- for (int from=1; from<=51; from++) {
- int[] too=ml.shiftTo(from);
- int[] toofrom=ml.shiftFrom(too[0]);
- System.out.println("ShiftFrom("+from+")=="+too[0]+" % "+too[1]+"\t+-+\tShiftTo("+too[0]+")=="+toofrom[0]+" % "+toofrom[1]);
- }*/
- System.out.print("Success?\n"); // if we get here - something must be working!
+ return fromShifts.size() == 1 && toShifts.size() == 1;
}
+
/**
+ * Returns the [start, end...] positions in the range mapped from, that are
+ * mapped to by part or all of the given begin-end of the range mapped to.
+ * Returns null if begin-end does not overlap any position mapped to.
*
- * @return a MapList whose From range is this maplist's To Range, and vice versa
+ * @param begin
+ * @param end
+ * @return
*/
- public MapList getInverse()
+ public int[] getOverlapsInFrom(final int begin, final int end)
+ {
+ int[] overlaps = MappingUtils.findOverlap(toShifts, begin, end);
+
+ return overlaps == null ? null : locateInFrom(overlaps[0], overlaps[1]);
+ }
+
+ /**
+ * Returns the [start, end...] positions in the range mapped to, that are
+ * mapped to by part or all of the given begin-end of the range mapped from.
+ * Returns null if begin-end does not overlap any position mapped from.
+ *
+ * @param begin
+ * @param end
+ * @return
+ */
+ public int[] getOverlapsInTo(final int begin, final int end)
{
- return new MapList(getToRanges(), getFromRanges(), getToRatio(), getFromRatio());
+ int[] overlaps = MappingUtils.findOverlap(fromShifts, begin, end);
+
+ return overlaps == null ? null : locateInTo(overlaps[0], overlaps[1]);
}
}