peak detection edge window is off by one resulting in wrong peaks identified

[christianschneebeli]

In detect_peaks() from peakdetection.py the edges are calculated as

    peakedges = np.concatenate((np.array([0]),
                             (np.where(np.diff(peaksx) > 1)[0]),
                             np.array([len(peaksx)])))

As a result, each edge (except the first) start one index too early and finish one index to early. cf:
peaksy[peakedges[i]:peakedges[i+1]].tolist()
Thus if for example the last y_value from the first edge is larger than all the y values in the second edge, it is detected as the peak from the second edge.
This would be the correct indices to slice the edges.

    peakedges = np.concatenate((np.array([0]),
                                (np.where(np.diff(peaksx) > 1)[0]+1),
                                np.array([len(peaksx)+1])))