Current version 6.3.3.

A new batch-wise processing procedure is added.

Download package at Sourceforge


Eearlier changes:

Version 6.2.4.

Two previously hidden parameters are brought to the wrapper level to facilitate more flexible data processing. The first is component.eliminate: in fitting mixture of bi-Gaussian (or Gaussian) model of an EIC, when a component accounts for a proportion of intensities less than this value, the component will be ignored. A higher value will suppress tiny peaks that happen to have the same m/z with a large peak. The second is moment.power:  the power parameter for data transformation when fitting the bi-Gaussian or Gaussian mixture model in an EIC. This parameter is to stablize the estimation of peak areas.

Version 6.2.2.

Corrected a bug in peak shape estimation. This bug is only relevent when the peak is in an EIC with small intensities far away from the peak.

Version 6.2.0.

Implemented a more efficient routine for the run filter.

Version 6.0.3.

Modified the code to load data. In R version 3.1, there is an issue loading netCDF files. Now it is corrected.

Version 6.0.1.

Major changes were made to the hybrid workflow, to avoid producing duplicate rows in the feature table.

Version 5.9.6.

Two new tables are included. One is a known metabolite table based on HMDB. The other is a table of potential adducts. A new function can produce the known feature table based on adducts of the user's choice.

Earlier versions of apLCMS assumed that m/z values are monotone increasing in each scan. In rare cases this is not true. The new version addressed this issue.

Version 5.9.2.

The major change in this version is we now rely on the mzR package from Bioconductor for data input. Now our package accepts all data formats that mzR allows, including .cdf, .mzXML, .mzML and others.


To ensure continuity, the commands in the package didn't change names. They still have "cdf" in their names. However they are not limited to NetCDF data format now.

Version 5.8.0.

The major change of this version is the parallelization of the retention time adjustment and feature alignment, and an improved aligning scheme making feature alignment in higher m/z range more reliable.


Version 5.6.0.

The major change of this version is the addition of the semi-supervised (hybrid) feature detection procedure. It is introduced as a wrapper function semi.sup(). It requires a data frame that records the known/historical features. Please see instruction page for details.

Version 5.4.0.

There are major changes in this version. 

(1) To adapt to non FT data, the automated m/z tolerance search at the single spectrum step was abandoned. The user's input of m/z tolerance is necessary. The user can use the machine's nominal accuracy level (ppm divided by 10^6). It is recommended to use 10^-5 for FTMS, and higher value for other machines.

(2) A new multi-level smoothing procedure is implemented to adaptively slice the profile into ion traces. 

(3) Parallel computing using multiple nodes on a single machine is implemented. It is based on SNOW. It requires packages from CRAN: doSNOW, iterators, foreach, and snow. 


Version 5.9.8.

HMDB 3.5 is incorporated.