sgibb / maldiquant Goto Github PK

• add arguments (default arguments, too)
• explain the need to adjust different arguments (e.g. halfWindowSize)

• new plotMsiSlice instead of plotImsSlice.

Use roxygen2 to create Rd files.

minNumber should be reduced to the maximal number of MassPeaks objects in the current label otherwise all peaks will removed:

library("MALDIquant")

p <- c(createMassPeaks(1:3, 1:3),
       createMassPeaks(1:2, 1:2),
       createMassPeaks(1, 1))

filterPeaks(p, minNumber=2, labels=c(1, 2, 2))

# [[1]]
# S4 class type            : MassPeaks
# Number of m/z values     : 0        
# Range of m/z values      : NA       
# Range of intensity values: NA       
# Range of snr values      : NA       
# Memory usage             : 1.234 KiB
# 
# [[2]]
# S4 class type            : MassPeaks
# Number of m/z values     : 1        
# Range of m/z values      : 1 - 1    
# Range of intensity values: 1 - 1    
# Range of snr values      : NA - NA  
# Memory usage             : 1.258 KiB
# 
# [[3]]
# S4 class type            : MassPeaks    
# Number of m/z values     : 1            
# Range of m/z values      : 1 - 1        
# Range of intensity values: 1e+00 - 1e+00
# Range of snr values      : NA - NA      
# Memory usage             : 1.258 KiB 

expected output:

# warning: reduce minNumber for label "1" because minNumber > n doesn't make sense!

# [[1]]
# S4 class type            : MassPeaks
# Number of m/z values     : 3        
# Range of m/z values      : 1 - 3    
# Range of intensity values: 1 - 3    
# Range of snr values      : NA - NA  
# Memory usage             : 1.281 KiB

Some raw files contain integer intensity values (e.g. fid files). Summing all this integers could cause an integer overflow.

totalIonCurrent(g[[1]])
# [1] NA
# Warning message:
# In sum(object@intensity, na.rm = TRUE) :
#   Integer overflow - use sum(as.numeric(.))

To reproduce:

library("MALDIquant")

s <- c(createMassPeaks(1:4, rep(1, 4), metaData=list(name=33)),
       createMassPeaks(1:4, rep(1, 4), metaData=list(name=33)),
       createMassPeaks(1:4, rep(1, 4), metaData=list(name=22)),
       createMassPeaks(1:4, rep(1, 4), metaData=list(name=22)),
       createMassPeaks(1:4, rep(1, 4), metaData=list(name=34)),
       createMassPeaks(1:4, rep(1, 4), metaData=list(name=34)))

f <- factor(rep(c(33, 22, 34), each=2))

p <- mergeMassPeaks(s, f)

sapply(p, function(x)metaData(x)$name)
## order is 22 33 44 instead of 33, 22, 34

Adapt to alignSpectra and new intensityMatrix (separate binPeaks call is not needed anymore).

Since fix #23 mergeMassSpectra throws a warning:

MassSpectrum created with missing/filtered mass values.

(which is not unusal in merging and should not confuse the user)

Seems to be a rounding problem:

> library("MALDIquant")
> createMassSpectrum(1:10, 1:10)
#S4 class type            : MassSpectrum
#Number of m/z values     : 10          
#Range of m/z values      : 1 - 10      
#Range of intensity values: 1 - 10      
> createMassSpectrum((1:10)+0.2, 1:10)
#S4 class type            : MassSpectrum
#Number of m/z values     : 10          
#Range of m/z values      : 1.2 - 10.2  
#Range of intensity values: 1 - 10      
#Warning message:
#In validityMethod(object) :
#  MassSpectrum created with missing/filtered mass values. This may compromise subsequent baseline correction and peak detection.

Merge Monoisotopic Peak Detection from MALDIquantTools into MALDIquant.

filterPeaks doesn't allow different minFrequency arguments for different labels. It only allows one minFrequency argument for all labels. That's only fine if all biological replicates have the same number of technical replicates.

Wrapper around transformIntensity.

R CMD check output:

File
LICENSE
is not mentioned in the DESCRIPTION file. 

MALDIquant is distributed under GPL (>=3) which is a supported license. We already use the License: GPL field in DESCRIPTION. That's why the LICENSE file could simply removed (it should be used only for unsupported licenses).

Currently the TIC is calculated by sum(spectrum@intensity) but this is only correct if all spectrum@mass values are equally spaced. For most spectra generated by Bruker devices this isn't true, because the difference between to mass values increases.
Something like (head(intensity, -1)+tail(intensity, -1))/2 * diff(mass) is needed.

In the future, pls explain

MALDI (11:47)
TOF (11:53)
TopHat (13:37)

in the Description

The representation argument in setClass was marked as deprecated (replaced by slots and contains).

Centroided or missing mass values break our window-based baseline correction and peak detection algorithms. That's why the constructor createMassSpectrum has to test for centroided or missing mass values and shows a warning if needed.

filterPeaks use floor to determine minimal number of peaks from a given minFrequency argument but it should not round it because this behavior is mostly unexpected.

To reproduce:
minFrequency=2/3;
2/3_3 == 2 (at least 2; mass == 1:4 for group 1);
2/3_4 == 2.6667 (=> at least 3 peaks (mass == 1:4) for group 2; but filterPeaks uses floor and we get mass == 1:5)

library("MALDIquant")

p <- c(createMassPeaks(1:3, 1:3),
       createMassPeaks(1:4, 1:4),
       createMassPeaks(1:5, 1:5),
       createMassPeaks(1:3, 1:3),
       createMassPeaks(1:4, 1:4),
       createMassPeaks(1:5, 1:5),
       createMassPeaks(1:6, 1:6))

sapply(filterPeaks(p, minFrequency=2/3, labels=c(rep(1, 3), rep(2, 4))), mass)

• replaced by .stopIfNotIsMass* (don't forget man/MALDIquant-internal.Rd)
• adapt MALDIquantTools

iplot uses some platform specific code, causes NOTE s on CRAN checks and the most important function getGraphicsEvent has some bugs e.g. https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=15208

see also: #13

• SNIP
• ConvexHull
• dilation/erosion
• localMaxima

• TIC
• Median
• Probabilistic Quotient Normalization (Huber/Dieterle)

Mark

• calibrate on matrices
• standardizeTotalIonCurrent on MassSpectrum objects
• totalIonCurrent<-

as deprecated.

Documentation

• calibrateIntensity (TIC, Median, PQN)
• add calibrateIntensity to workflow example

R 3.0.0 will add LongVector support:
http://stat.ethz.ch/R-manual/R-devel/library/base/html/LongVectors.html

Length of mass and intensity are possibly no integers anymore. That's why some C code has to be rewritten. All .C-calls have to be replaced by .Call (#14).

( Low priority because who want's to (or could) create spectra with more than 2^31-1 data points?:open_mouth:)

• SNIP
• ConvexHull
• dilation/erosion
• localMaxima

Don't forget MALDIquant-internal.Rd.

averageMassSpectra will replace current mergeMassSpectra (should be marked as deprecated or should really merges spectra) and will work on a mass range that is covered by all spectra.

To reproduce:

library("MALDIquant")                                                                                         
s <- createMassSpectrum(1:2, 1:2)                                                                             
estimateBaseline(s, method="Median")

Output:

Warning message:
In runmed(y, k = (2 * halfWindowSize + 1)) :
'k' is bigger than 'n'!  Changing 'k' to 1

Don't forget NAMESPACE and man page.

To reproduce:

library("MALDIquant")
data(fiedler2009subset)
MALDIquant:::.doByLabels(fiedler2009subset[1:4], labels=1:4, function(x)x[[1]])

Error message:

Error in `split<-.default`(`*tmp*`, f, drop = drop, value = value) :
  object of type 'S4' is not subsettable

It should be possible to use TIC only on a user-defined range.
(If no range is given the largest overlap should be chosen.)

• general
• warping/binning + intensity calibration
• biomarker/classification; see Analysis of Fiedler et al. 2009 using MALDIquant in the MALDIquantExamples package.
• phylogentic; see Bacterial Species Determination using MALDIquant in the MALDIquantExamples package.

Don't forget NAMESPACE and MALDIquant-internal.Rd.

Package in Depends field not imported from: ‘methods’
These packages needs to imported from for the case when
this namespace is loaded but not attached.
See the information on DESCRIPTION files in the chapter ‘Creating R
packages’ of the ‘Writing R Extensions’ manual.

cleanup and move tests from ../tests to inst/tests/

Use mclapply and mcmapply to allow parallel execution at least on unix based platforms.

Would you consider depending on ProtGenerics and use the mass and intensity generics that are defined there? They are defined as follows

setGeneric("intensity", function(object, ...) standardGeneric("intensity"))
setGeneric("mass", function(object, ...) standardGeneric("mass"))

testthat 0.8 comes with a new recommended structure for storing your tests. To
better meet CRAN recommended practices, testthat now recommend that you to put
your tests in tests/testthat, instead of inst/test (this makes it
possible for users to choose whether or not to install tests). With this
new structure, you'll need to use test_check() instead of test_packages()
in the test file (usually tests/testthat.R) that runs all testthat unit
tests.