
### MBT added 12-2-08

userError<-function (msg) 
{
    stop(msg, call. = FALSE)
}
### MBT end


noiseEstmtr2 <- function(spect)
{

#----------------------------------------------------
# Estimate noise level (STD) for TOF record (assuming constant peak width)
#
# INPUT:
# spect - spectrum signal (resampled or range of constant peak width);
#
# OUTPUT:
# ns - noise estimate assuming Gaussian (random) noise;
#
# USAGE:
# ns <- noiseEstmtr2 (spect);
# Dependency: none
#

 
## Part 1  Noise Level Estimation

len <-length(spect); 

#A <- spect[1:(len-1)]; B <- spect[2:len]; d1 <- A - B; 
d1 <- -diff(spect,1,1); 

spect <- sort(d1); 
#print(length(spect)) # debug
tstart <- min(which(spect>0))-round(0.05*len);
tend <- tstart+round(0.1*len);
#print(c(tstart,tend)) #debug
spect <- spect[tstart:tend];
tt <- 1:length(spect); #tt <- t(tt);
linef <- lsfit(tt, spect, intercept  = TRUE); # line(tt,spect);  
fcoefs <- coef(linef)
ns <- len*fcoefs[2]/3.55; 

return(ns);
}


trivPP_call <- function(tofList, snr, naml, HWFM, fwl, stPk, dataStart)
{
###----------------------------------------------------
### Find time position of the pivot peaks above SNR 
### input signal
###
### INPUT:
### tofList -  list of observed sampled discrete signals (rows);
### snr - signal-to-noise ration for peak detection
### naml - noise amplitude list (for tofList);
### HWFM  half-width at half-maximum for the model signal wavelet
### fwl - filter wavelet length applied for signal processing
### stPk - starting point (tof-spectrum index) for peak detection (e.g., after matrix hump)
### dataStart - defines what is the absolute start location for the input time points
###
###
### OUTPUT:
### peakList - time point numbers for pivots;
###


spectraName <- names(tofList)

Ns <- length(tofList)

if ( snr <= 0 )
{
	userError( "Signal-to-Noise ratio must be positive." );
}


if ( HWFM < 1 )
{
	userError( "Invalid Peak Half-Width; it must be at least 1.0" );
}

pivLength=0;

## define the results list
peakList <- list()

for (i in 1:Ns){
 spec <- tofList[[i]]
 ## zero-fill for NA values
 spec[is.na(spec)] <- 0 
 peakTime <- spec * 0;
 peakAmps <- peakTime
 timeUncerts <- peakTime
 ampUncerts <- peakTime
 
 Nt <- length(spec)
 ampN <- as.numeric(naml[[i]]);
 stdNoise <- ampN/3;
	
     #print(c(length(ampN),Nt)) #debug
 if (Nt != length(ampN)) {
	userError("Noise amplitude vector is not the same length as signal")
 }
    	qin <- 1:Nt
	k<-qin; #qin[spec/ampN >= snr] ## use noise amplitude 11/08/07 << REV
	if ( length( k ) >= 3 )
	{     ## Step 1: find all local maxima and minima
		x<-diff(spec[k],1,1)
		xl<-x[1:length(x)-1]
		xr<-x[2:length(x)]
		nrtp <- (xl>0 & xr<=0) | (xl==0 & xr<0);
		nrbt<- (xl<0 & xr >=0) | (xl==0 & xr>0)
		xx <- 1:length(x);
		nb <- c(1, xx[nrbt]+1); ## tentative minimum
		nt <- xx[nrtp]+1; ## tentative maximum
		
		## Step 2: Coalesce rising plateaus to find the right-most one
		curt <- 1; curb <- 1; peak <- c(0); bots <- c(1); 
		numtop <- 1; numbot <- 1;
		while ((curt < length(nt)) & (curb < length(nb))) {
			while ((nb[curb]<nt[curt]) & (curb < length(nb))) {
			curb <- curb+1;
			}
		bots[numbot]<-nb[curb];
		numbot<-numbot+1;
			while ((nt[curt]<nb[curb]) & (curt < length(nt))) {
			curt <- curt+1;
			}
		peak[numtop]<-nt[curt];
		numtop <- numtop+1;
		}
		rm(numtop, numbot, curt, curb, nb, nt, nrbt, nrtp)
		leftMin <- c(1); rightMin <- c(1);
		for (i1 in 2:length(peak)) {
		temp <- peak[i1-1]:peak[i1];
		leftMin[i1] <- max(temp[spec[k[temp]]==min(spec[k[temp]])]);
		rightMin[i1] <- min(temp[spec[k[temp]]==min(spec[k[temp]])]);
		}
		rightMin <- c(rightMin[2:length(rightMin)], length(x));
 
		con1 <- abs(spec[k[leftMin]]-spec[k[peak]])/ampN[k[peak]]; ## << REV
		con2 <- abs(spec[k[rightMin]]-spec[k[peak]])/ampN[k[peak]]; ## << REV for ampN vector
		
		peak <- peak[(con1 >= snr) | (con2 >= snr)];
		peak <- peak[spec[k[peak]]/ampN[k[peak]] >= snr];

		if ( length( peak ) > 0 )
		{
			pivots<-k[peak]
		} else
		{
			pivots <- 0;
		}
	} else
	{
		if ( length( k ) > 0 )
		{
			pivots <- k[spec==max(spec[k])];
		} else
		{
			pivots <- 0;
		}
	}	
	## use only peaks no closer than 2*HWFM to the origin (to estimate uncertainties)
	pivots<-pivots[(pivots>2*HWFM)]   ## bug fix for 6086, Dasha 06/22
	pivots <- pivots[which(pivots>stPk & pivots < (length(spec)- ceiling(fwl/2)))]; # trustworthy range

	numberPeaks <- length(pivots)
#print(length(pivots)) #debug
	if ( numberPeaks > 0 )  ## REV << check that stdNoise is vector or add ampN/3?
	{
		peakTime[1:numberPeaks]<-pivots
#print(peakTime) #debug
		peakAmps[1:numberPeaks]<- spec[peakTime[1:numberPeaks]];
		ampUncerts[1:numberPeaks]<-peakTime[1:numberPeaks]*0+ampN[pivots];
		snrPeaks<-peakAmps[1:numberPeaks]/stdNoise[pivots];
	
		for (j in 1:numberPeaks){
			if ((snrPeaks[j]>2) & (peakTime[j]>2*HWFM)){
				y<-1.0/(1.0-2/snrPeaks[j]);
				tempd <- spec[(peakTime[j]-2*HWFM):peakTime[j]]
				timeUncerts[j]<-length(tempd[tempd>=tempd[length(tempd)]/y])
			}else{
				timeUncerts[j]<-HWFM;
			}
		}
	##print("assign_info_list")
		info <- list( 
			"Positions" = peakTime[1:numberPeaks]+dataStart-1,
			"Intensities" = peakAmps[1:numberPeaks],
			"PositionUncertainties" = timeUncerts[1:numberPeaks],
			"IntensityUncertainties" = ampUncerts[1:numberPeaks] )
#print(info$Positions) #debug
	} else
	{       print("no_peaks")
		info <- list( 
			"Positions" = NULL,
			"Intensities" = NULL,
			"PositionUncertainties" = NULL,
			"IntensityUncertainties" = NULL )
	}

	## append results to the list
  	peakList[i] <- list( info )  # allow assignment for repeated names 02/11/10
#print(peakList[[i]]$Positions) #debug

}
names(peakList) <- spectraName;
return(peakList);
#tofList <- NULL;
}


peakAlignAv<-function(peaks,tofList,noiseL,HWFM){		

###----------------------------------------------------------------- 
### Function peakalignAv aligns peaks from tofList spectra with peaks found
###   in averaged spectrum 
###
### INPUT:
### 	peaks - peak positions found in the averaged spectrum;
###	tofList - list of TOF spectra in the same domain (down-sampled or original) as "peaks"
### 	HWFM - peak half-width (window for alignment) -- constant for down-sampled data;
###
### OUTPUT:
### 	alignedPeakList
###
###------------------------------------------------------------------

  alignedPeakList <- list();
  alignedPeakList$peaks <- peaks;
  alignedPeakList$data <- list();

  specName<-names(tofList)
  numSpectra<-length(tofList)
  numPeaks<- length(peaks) 
  peakInt<-array(0,numPeaks) 
  peakShifts<-peakInt 

  zeroVector<-rep(0,numPeaks)

## align

for (i in 1:numSpectra) {
    spectra <- tofList[[specName[i]]]
    nam <- noiseL[[i]]

    for (j in 1:numPeaks) { 	
        w <- spectra[(peaks[j]-HWFM):(peaks[j]+HWFM)];      
       mw<-max(w) 
      imw <- which.max(w)
      x <- diff(w,1,1);		# first difference
	xEnd<-length(x)
       xl <- x[1:xEnd-1]	# left side
       xr <- x[2:xEnd]		# right side
       neartop <- (xl > 0 & xr <= 0) | (xl == 0 & xr < 0)     
      xx <- 1:xEnd
      nt <- xx[neartop]+1 		# tentative maximum     
      if(length(nt)>0) {       
          if(length(nt)>1) { 
            y<-max(w[nt])
            iy<-which.max(w[nt])
           peakInt[j]<- y  
           peakShifts[j]<-nt[iy]-HWFM-1 
           }
          else {   
              peakInt[j]<-w[nt]
              peakShifts[j]<-nt-HWFM-1                
          }    
     } else {
            if(abs(imw-HWFM-1) < HWFM) {  
             peakInt[j] <- mw 
                peakShifts[j] <- imw-HWFM-1 
          }  else {
                 peakInt[j]<-spectra[peaks[j]] 
                 peakShifts[j]<-0  
            } 
      }       
    }     
 
peakInt<-as.numeric(peakInt)
uncertInt <- nam[peaks+peakShifts] # local noise estimate at peak position

## Construct alignedPeakList

 	alignedPeakData <- list("Intensities" = as.vector( peakInt ),
  		"PositionShifts" = as.vector(peakShifts),
  			"PositionUncertainties" = zeroVector + max(abs(peakShifts)),
  				"IntensityUncertainties" = uncertInt )
  
  	alignedPeakList$data[i] <- list( alignedPeakData ); # allow assignment for repeated "names"
  }
  names(alignedPeakList$data) <- specName;
  return(alignedPeakList)
}


#########GLobal alignment routines below########
######### Pivot peak finder ############
pivPeak <- function(refsig, thres, phw) {

###----------------------------------------------------
### Find time position of the pivot peaks above threshold 
### input signal
###
### INPUT:
### refsig - observed sampled discrete reference signal nromalized to its maximum = 1;
### thres - intensity threshold for pivot peaks in respect to maximum (e.g. 0.1);
### phw - peak half-width in the pivot detection range
###
### OUTPUT:
### pivots - time point numbers for pivots;
###

qin<-1:length(refsig);
k<-qin[refsig>thres];
x<-diff(refsig[k],1,1)
xl<-x[1:length(x)-1]
xr<-x[2:length(x)]
p<-2:length(k)-1;
p1<-p[(xl>0 & xr<=0) | (xl==0 & xr<0)]
pivots<-k[p1]
n1 <- which(diff(pivots)<phw);
loopn <- 0;
while (length(n1)>0 & loopn < 10){  # combine pivots closer than HW (to get 1/peak)
	loopn <- loopn + 1;
	#print(loopn)
	pcomp <- cbind(pivots[n1],pivots[n1+1]);
	comp <- c(pivots[n1],pivots[n1+1])
	dpiv <- setdiff(pivots,comp)
	pcomp1 <- pcomp[,1];
	d1 <- dim(pcomp)[1]
	for (jj in 1:d1){  # select pivot of max signal for closer than HW pivots
    	pcomp1[jj] <- pcomp[jj,which(refsig[pcomp[jj,]]==max(refsig[pcomp[jj,]]))]
    	}
	pivots <- sort(unique(c(pcomp1,dpiv)))
	n1 <- which(diff(pivots)<phw);
	rm(pcomp,comp,dpiv, pcomp1,d1,jj)
}
rm(qin,k,x,xl,xr,p,p1,n1,loopn)
return(as.vector(pivots))

}

########## Local jitter detection by correlation (for global alignment) ##########
locjitdet <- function(fs, ps, sw, startS, endS, spectraNames){

###-----------------------------------------------------------
### Compute jitter  and triggering error in time points for the set of signals
### at the pivit peak positions.
### Jitter is measured by the shifts of the maxima for the local
### cross-correlation function in the vicinity of a peak.
### Jitter is detected in resspect to the FIRST (reference) signal in the set.
###
### INPUT:
### fs - matrix of sampled discrete signals (columns - intensities)
### ps - peak set for fs precomputed eg. with "pivpiks"
### sw - maximum lag for cross-correlation (shift window ~ of twice
### peak sigma, estimated eg. with "peakwidth_pf");
###
### OUTPUT:
### out$pjit = fsj (i,j) - jitter in point numbers for the corresponding peak
### "i" in the signal "j" as measured by the local cross-correlation to the reference signal;
### out$trerr = avshift(j) - average triggering error for the j-th record
###-----------------------------------------------------------

### "sw" is a window for local correlation function (maximum lag)
### sw=10 is recommended for SELDI data (focused from 2 to 15 kDa)

	cw<-3*sw; # is a half-window in a spectrum for the local X-corr calculation
	ps<-as.vector(ps)
	
	l0<-length(ps)

	### remove peaks for which correlation cannot be calculated with the required window
	maxpoint<-eval(nrow(fs))
	ps[ps<(1+cw)]<-0               ### remove peaks closer than correlation window to the record's zero
	ps[ps>(maxpoint-eval(cw)-1)]<-0 ### remove peaks closer than correlation window to the record's end
	t<-ps[ps!=0]
	ps<-t
	
	Nr<-length(ps);
	if ( Nr < 3 )
	{
		userError( "Could not find at least 3 peaks in the reference spectrum." );
	}

	Nc<-ncol(fs); ### if we have multiple spectra

	if (Nr<l0) { print('WARNING: less peaks used -- some peaks outside allowed correlation window')}

	fsj<-matrix(rep(0,Nr*Nc),nrow=Nr,ncol=Nc);  ### initialize jitter table for the peaks
	msj<-fsj; ### initialize maximum correlation coefficient table for the peaks

	enhstart<-eval(cw-floor(sw/2)) ### peak enhancement range for the reference signal
	enhend<-eval(cw+floor(sw/2))
	refsig <- fs[,1]/max(fs[startS:endS,1]);
	lmax <- trunc(length(startS:endS)/2);
	#lmax <- length(startS:endS)-1;
	if (Nr>0)
	{
		for (j in 2:Nc)
		{   
			### calculate jitter for all peaks
			nsigJ <- fs[,j]/max(fs[startS:endS,j]);
			cfg <- ccf(refsig[startS:endS], nsigJ[startS:endS], lag.max=lmax, type="correlation", plot=FALSE);
			cw <- 3*abs(cfg$lag[cfg$acf==max(cfg$acf)]) + 3*sw;
			
			cwStart <- cw + 1;
			cwEnd <- length( refsig ) - cw;
			
			peaksWindowStart <- ps[ ps < cwStart ];
			peaksWindowEnd <- ps[ ps > cwEnd ];
			
			windowStartError <- length( peaksWindowStart ) > 0;
			windowEndError <- length( peaksWindowEnd ) > 0;
			
			if ( windowStartError || windowEndError )
			{
				msg <- paste( "\nWhile aligning spectrum '", spectraNames[j], "' to reference spectrum '", spectraNames[1], "':\n\n", sep = "" );
				checkRefSpecMsg <- "\nPlease check the reference spectrum.";
				if ( cwStart >= cwEnd )
				{
					userError( paste( msg, "For all pivot peaks, the correlation window goes beyond the range of the spectrum data.", checkRefSpecMsg, sep = "" ) );
				}
				
				msg <- paste( msg, "For some pivot peaks, the correlation window goes beyond the range of the spectrum data.", sep = "" );
				msg <- paste( msg, checkRefSpecMsg, sep = "" );
				msg <- paste( msg, "\n\nThe peaks that caused this problem are ", sep = "" );
				
				correctionStart <- "\nIf the reference spectrum is good, then adjusting ";
				correctionEnd <- " will exclude these peaks and should resolve this error.";
				
				if ( windowStartError && windowEndError )
				{
					msg <- paste( msg, "<= ", peaksWindowStart[ length( peaksWindowStart ) ], " and >= ", peaksWindowEnd[ 1 ], ".", sep = "" );
					msg <- paste( msg, correctionStart, "'Start Time' to ", cwStart, " and 'End Time' to ", cwEnd, correctionEnd, sep = "" );
				} else if ( windowStartError )
				{
					msg <- paste( msg, "<= ", peaksWindowStart[ length( peaksWindowStart ) ], ".", sep = "" );
					msg <- paste( msg, correctionStart, "'Start Time' to ", cwEnd, correctionEnd, sep = "" );
				} else
				{
					msg <- paste( msg, ">= ", peaksWindowEnd[ 1 ], ".", sep = "" );
					msg <- paste( msg, correctionStart, "'End Time' to ", cwEnd, correctionEnd, sep = "" );
				}
				
				userError( msg );
			}
			
			for (i in 1:Nr)
			{
				startp<-eval(ps[i]-cw)
				endp<-eval(ps[i]+cw)
				
				refc<-refsig[startp:endp];

				fsc<-nsigJ[startp:endp];

				cf<-ccf(refc,fsc,lag.max=trunc(cw/2), type="correlation", plot=FALSE);
				mm<-max(cf$acf);
			    fsj[i,j]<-cf$lag[cf$acf==max(cf$acf)];

			    #if ((fsj[i,j]==trunc(cw/2)) | (fsj[i,j]==-trunc(cw/2))) {fsj[i,j]=0;} 
		    	msj[i,j]<-mm;
			}
 		}
	} else {print('WARNING: EMPTY JITTER MATRIX -- all peaks outside allowed correlation window')}

	avshift<-as.vector(rep(0,Nc-1))

	for (jj in 1:(Nc-1))
	{
		signjj<-sign(fsj[,jj+1])
		lneg<-force(length(signjj[signjj<0]))
		lpos<-force(length(signjj[signjj>0]))
		lzer<-force(length(signjj[signjj==0]))
		votesign<-max(c(lneg, lpos, lzer))

		avneg<-sum(fsj[signjj<0,jj+1])
		if (lneg>0){avneg<-avneg/lneg}

		avpos<-sum(fsj[signjj>0,jj+1])
		if (lpos>0){avpos<-avpos/lpos}

		avzer=sum(fsj[signjj==0,jj+1])
		if (lzer>0){avzer<-avzer/lzer}

		avshift[jj]<-avneg*(lneg==votesign)+avpos*(lpos==votesign)+avzer*(lzer==votesign)
	}

	jitout <- list(pjit=fsj[,], trerr=avshift)

	return(as.list(jitout))
	#return(as.vector(fsj));
}

################## Global Alignment Function ##############
globalAlign<- function(tof,tofList, tofListMetaData, HWFM,noiseThres,startS,endS, Appli, Detect) {

### INPUT:
### tof - baseline subtracted and smoothed tofList (to detect global shifts)
### tofList - unprocessed data (to apply detected shifts)
### tofListMetaData - "offset" and "scale" are used and /or updated
### HWFM - peak width in the range of pivot detection
### noiseThres - fraction of maximum signal threshold for pivot detection
### startS, endS - TOF range (indices) for pivot detection
### Appli - apply correction to the tofList
### Detect - auto-detect correction for the data

Nr <- length(tofList)
spectraName <- names(tofList)
Nc <- 0
for(i in 1:Nr) {
  	len <- length(tof[[i]])
  	if (len > Nc) {
    	Nc <- len
  	}
}

sigtm <- matrix(0, nrow=Nr, ncol=Nc)
sigt0 <- sigtm

digitizerDwell <- as.double(tofListMetaData[1,"instrumentSpecificSettings.timeDelta"]) ;
for (i in 1: Nr) {
    	sigt <- tof[[i]]
    	sig0 <- tofList[[i]]
    	l1 <- length(sigt)
    	l0 <- length(sig0)
    	#print(c(l1,l0)) 
    	sigtm[i, 1:l1] <- sigt;
    	sigt0[i, 1:l0] <- sig0;
    	sigtm[i, is.na(sigt)]<-0  ## zero-fill for NA values 06/22, Dasha
    	sigt0[i, is.na(sig0)]<-0 
	if ( digitizerDwell != as.double(tofListMetaData[ i, "instrumentSpecificSettings.timeDelta" ]) )
	{
		userError( "Spectra do not all have the same digitizer frequency.  Please choose spectra that all have the same experimental settings." );
	}
}

if(Detect){
	sw <- 2 * HWFM
	if ( sw < 2 )
	{
		userError( "The 'Peak Half-Width' must be at least one time tick." );
	}

	if ( noiseThres <= 0 || noiseThres >= 1 )
	{
		userError( "'Threshold Intensity' must be between 0 and 1." );
	}

	### specify the time range to look for pivot peaks

	### try to transfer the list to a matrix and error out if the length are different

	
	dataStart<-as.integer(tofListMetaData[1, "timeOfFlightData.start" ]);
	if ( startS < dataStart )
	{
		userError( "'Start Time' must be between 'dataStart' and the longest spectrum length." );
	}

	if ( endS > Nc )
	{
		userError( "'End Time' must be less than the longest spectrum length." );
	}

	if ( endS <= ( startS + 3 * sw ) )
	{
		userError( "The specified time range ('End Time' - 'Start Time') must be greater than 18 peak half-widths in order to contain at least 3 pivot peaks in the reference spectrum." );
	}

	pivPeaks<- pivPeak(sigtm[1,startS:endS]/max(sigtm[1,startS:endS]), noiseThres, HWFM);
	pivPeaks <- pivPeaks + startS;
	print("Found pivot peaks for global alignment:")
	print(length(pivPeaks))
	#print(pivPeaks)
	if ( length( pivPeaks ) < 3 )
	{
		userError( "Could not find at least 3 peaks in the reference spectrum." );
	}

	locJit <-locjitdet(t(sigtm), pivPeaks, sw, startS, endS, names(tofList));

	pj <- locJit$pjit;  ### pj[i,j] shows jitter for peak "i" in signal "j"
	#print(pj) #debug
	#pjSlope <- rep(0,dim(pj)[2]); pjIntercept <- pjSlope;
} else { pivPeaks <- c(1,2)  # no auto-detection is performed
	pj <-rbind(rep(0,Nr),rep(0,Nr))
}

	t0 <- 1:Nc ## original time axis
	sapl <- 0;

	for (j in 2:(dim(pj)[2])){ ## calculate linear fit coefficients for jitter in "j"-th record
 		ldf <- length(which(pj[,j]!=0)); lndf <- length(pivPeaks)-ldf; 
 		if ( (ldf > 3) & ((ldf-lndf) > 2)) { ## enough data to fit to a line
			z<-line(pivPeaks, pj[,j]);
			pjIntercept<-coef(z)[1];
			pjSlope<-(1.0+coef(z)[2]);
		} else { pjSlope <- as.numeric(tofListMetaData[ j, "timeOfFlightData.scale" ]); 
			pjIntercept <- as.numeric(tofListMetaData[ j, "timeOfFlightData.offset" ]);
		} ## input meta-data values
		#print(c(pjIntercept, pjSlope))  #debug

		if( Detect & !Appli){
			tofListMetaData[ j, "timeOfFlightData.offset" ] <- pjIntercept
			tofListMetaData[ j, "timeOfFlightData.scale" ] <- pjSlope
		}

		if ( Appli ) { ## apply new time scale
        		if (pjSlope!=1 | pjIntercept!=0) {
				tsj<-pjSlope*t0+pjIntercept
				intrp <- approx(tsj, sigt0[j,], t0)  ## appli to "original" (no BS) spectrum
				outSpectrum <-as.vector(intrp$y);
				inputSpectrum <- tofList[[j]]
				#print(c(length(inputSpectrum),length(outSpectrum))) #debug
				outSpectrum <- outSpectrum[1:length(inputSpectrum)]
				outSpectrum[is.na(outSpectrum)] <- inputSpectrum[is.na(outSpectrum)]
				tofList[spectraName[j]] <- list(outSpectrum)
        			sapl <- c(sapl,j);
        		}
        		tofListMetaData[ j, "timeOfFlightData.offset" ] <- pjIntercept
			tofListMetaData[ j, "timeOfFlightData.scale" ] <- pjSlope
        
		} 
 		#print(tofListMetaData[length(tofList), "timeOfFlightData.offset" ])
	}
	if (length(sapl)>1){
		sapl <- sapl[2:length(sapl)]; # spectra with applied global shift 
		print("####    Pre-aligned spectra:")
		print(sapl)
	} else {
		print("####    None of the spectra needed/required pre-alignment    ####")
	}
aOut <- list(tList=tofList, tMeta=tofListMetaData)
rm(sigtm, sigt0, sigt, sig0)
return(as.list(aOut))
}

shiftCorrect <- function(tofList)
{
return(tofList); ## noise thresholds depending on filter model <<EDIT
}


