Implements the isotope wavelet feature finder. More...

#include <OpenMS/TRANSFORMATIONS/FEATUREFINDER/FeatureFinderAlgorithmIsotopeWavelet.h>

Inheritance diagram for FeatureFinderAlgorithmIsotopeWavelet< PeakType, FeatureType >:

Classes
struct	BoxElement
	Internally used data structure for the sweep line algorithm. More...
Public Types
typedef FeatureFinderAlgorithm < PeakType, FeatureType >	Base
Public Member Functions
	FeatureFinderAlgorithmIsotopeWavelet ()
	Default Constructor.
virtual	~FeatureFinderAlgorithmIsotopeWavelet ()
	Destructor.
IsotopeWaveletTransform < PeakType >::TransSpectrum *	prepareHRDataCuda (const UInt i, IsotopeWaveletTransform< PeakType > *iwt)
MSSpectrum< PeakType > *	createHRData (const UInt i)
void	run ()
	The working horse of this class.
Static Public Member Functions
static const String	getProductName ()
static FeatureFinderAlgorithm < PeakType, FeatureType > *	create ()
Protected Types
typedef std::map< UInt, BoxElement >	Box
	Key: RT (index), value: BoxElement.
Protected Member Functions
void	updateMembers_ ()
	This method is used to update extra member variables at the end of the setParameters() method.
Protected Attributes
UInt	max_charge_
	The maximal charge state we will consider.
DoubleReal	intensity_threshold_
	The only parameter of the isotope wavelet.
UInt	RT_votes_cutoff_
UInt	real_RT_votes_cutoff_
UInt	RT_interleave_
	The number of subsequent scans a pattern must cover in order to be considered as signal.
String	use_gpus_
String	intensity_type_
bool	use_tbb_
bool	use_cuda_
bool	check_PPMs_
bool	hr_data_
std::vector< UInt >	gpu_ids_
	A list of all GPU devices that can be used.
Int	progress_counter_

Detailed Description

template<typename PeakType, typename FeatureType>
class OpenMS::FeatureFinderAlgorithmIsotopeWavelet< PeakType, FeatureType >

Implements the isotope wavelet feature finder.

The FeatureFinderAlgorithmIsotopeWavelet class has been designed for finding features in 1D or 2D MS data sets using the isotope wavelet. In the case of two dimensional data, the class provides additionally the sweep line algorithm. Please note that the algorithm implemented here is only marginally related to the algorithm presented in Schulz-Trieglaff et al. (2007, 2008), as no fitting procedure is applied anymore after the wavelet-based seeding step. The wavelet has been designed to extract even very lowly-abundant features (see Hussong et. al (2007, 2009)), usually featuring a very low signal-to-noise ratio. The wavelet in its current implementation is not able to resolve overlapping patterns (see also Hussong et. al (2009)) and slightly shifts masses to the right due to the construction of the wavelet.

Parameters of this class are:

Name	Type	Default	Restrictions	Description
max_charge	int	3	min: 1	The maximal charge state to be considered.
intensity_threshold	float	-1		The final threshold t' is build upon the formula: t' = av+t*sd, where t is the intensity_threshold, av the average intensity within the wavelet transformed signal and sd the standard deviation of the transform. If you set intensity_threshold=-1, t' will be zero. As the 'optimal' value for this parameter is highly data dependent, we would recommend to start with -1, which will also extract features with very low signal-to-noise ratio. Subsequently, one might increase the threshold to find an optimized trade-off between false positives and true positives. Depending on the dynamic range of your spectra, suitable value ranges include: -1, [0:10], and if your data features even very high intensity values, t can also adopt values up to around 30. Please note that this parameter is not of an integer type, s.t. you can also use t:=0.1, e.g.
intensity_type	string	ref	ref, trans, corrected	Determines the intensity type returned for the identified features. 'ref' (default) returns the sum of the intensities of each isotopic peak within an isotope pattern. 'trans' refers to the intensity of the monoisotopic peak within the wavelet transform. 'corrected' refers also to the transformed intensity with an attempt to remove the effects of the convolution. While the latter ones might be preferable for qualitative analyses, 'ref' might be the best option to obtain quantitative results. Please note that intensity values might be spoiled (in particular for the option 'ref'), as soon as patterns overlap (see also the explanations given in the class documentation of FeatureFinderAlgorihtmIsotopeWavelet).
check_ppm	string	false	true, false	Enables/disables a ppm test vs. the averagine model, i.e. potential peptide masses are checked for plausibility. In addition, a heuristic correcting potential mass shifts induced by the wavelet is applied.
hr_data	string	false	true, false	Must be true in case of high-resolution data, i.e. for spectra featuring large m/z-gaps (present in FTICR and Orbitrap data, e.g.). Please check a single MS scan out of your recording, if you are unsure.
sweep_line:rt_votes_cutoff	int	5	min: 0	Defines the minimum number of subsequent scans where a pattern must occur to be considered as a feature.
sweep_line:rt_interleave	int	1	min: 0	Defines the maximum number of scans (w.r.t. rt_votes_cutoff) where an expected pattern is missing. There is usually no reason to change the default value.