Method getReadCounts

src/rnabloom/util/PafUtils.java:387–477

@param pafPath input PAF file @param targets set of target names @param tolerance minimum extension required for bridging partners @return map of targets and their read counts @throws IOException

(String pafPath, Set<String> targets, int tolerance)

     * @throws IOException 
     */
    public static HashMap<String, Float> getReadCounts(String pafPath, Set<String> targets, int tolerance) throws IOException {
        HashMap<String, Float> counts = new HashMap<>(targets.size());
        
        PafReader reader = new PafReader(pafPath);
        
        String prevName = null;
        int largestOverlapSize = 0;
        Interval largestOverlap = null;
        String bestTarget = null;
        HashMap<String, Interval> targetNameQueryIntervalMap = new HashMap<>();
        for (PafRecord r = new PafRecord(); reader.hasNext();) {
            reader.next(r);
            
            if (targets.contains(r.tName)) {
                if (!r.qName.equals(prevName)) {
                    if (prevName != null) {
                        float increment = 1f;
                        targetNameQueryIntervalMap.remove(bestTarget);

                        String partner = getBestPartner(targetNameQueryIntervalMap, largestOverlap, tolerance);
                        if (partner != null) {
                            increment = 0.5f;
                            
                            // TODO: Store number of reads bridging two targets
                            if (counts.containsKey(partner)) {
                                float c = counts.get(partner);
                                counts.put(partner, c + increment);
                            }
                            else {
                                counts.put(partner, increment);
                            }
                        }
                        
                        if (counts.containsKey(bestTarget)) {
                            float c = counts.get(bestTarget);
                            counts.put(bestTarget, c + increment);
                        }
                        else {
                            counts.put(bestTarget, increment);
                        }
                        
                        targetNameQueryIntervalMap.clear();
                    }
                    
                    prevName = r.qName;
                    largestOverlapSize = 0;
                    bestTarget = null;
                }
                
                Interval i = new Interval(r.qStart, r.qEnd);
                targetNameQueryIntervalMap.put(r.tName, i);
                
                int overlapSize = r.qEnd - r.qStart;
                if (overlapSize > largestOverlapSize) {
                    largestOverlapSize = overlapSize;
                    bestTarget = r.tName;
                    largestOverlap = i;
                }