merge.Results: merge.Results

In yufree/xMSanalyzer: xMSanalyzer: R package for data analysis, comparison, and annotation of metabolomics data.

Description

Function that merges results from different parameter settings.

Usage

merge.Results(dataA, dataB, feat.eval.A, feat.eval.B, max.mz.diff = 15, 
    max.rt.diff = 300, merge.eval.pvalue = 0.05, alignment.tool = "apLCMS", 
    numnodes = 1, mult.test.cor = FALSE, mergecorthresh = 0.7, missingvalue = 0)

Arguments

dataA	feature alignment output matrix from apLCMS or XCMS with intensities at parameter settings P1
dataB	feature alignment output matrix from apLCMS or XCMS with intensities at parameter settings P2
feat.eval.A	feature evaluations results from evaluate.Features for results at parameter settings P1
feat.eval.B	feature evaluations results from evaluate.Features for results at parameter settings P2
max.mz.diff	+/- mz tolerance in ppm for feature matching
max.rt.diff	retention time tolerance for feature matching
merge.eval.pvalue	Threshold for defining signifcance level of the paired t-test or the Pearson correlation during the merge stage in xMSanalyzer. The p-value is used to determine whether two features with same m/z and retention time have identical intensity profiles.
mergecorthresh	Correlation threshold to be used during the merge stage in xMSanalyzer to determine whether two features with same m/z and retention time have identical intensity profiles.
alignment.tool	Name of the feature alignment tool eg: "apLCMS" or "XCMS"
numnodes	Number of computing nodes to use. Default: 1
mult.test.cor	Should Bonferroni multiple testing correction method be applied for comparing intensities of overlapping m/z? Default: FALSE
missingvalue	How are missing values represented. eg: 0 or NA

Details

We use a four-step process to merge features from different parameter settings. In step one, features detected at settings P1 and P2 are combined into one list. In step two, features are grouped by a user-defined m/z tolerance (5 ppm is appropriate for high resolution MS but may not be suitable for lower resolution instruments; for the LTQ-FT/MS, examination of m/z tolerance shows little difference between 5 and 10 ppm). In step three, features are further sub-grouped based on a user-defined retention time tolerance. Users are recommended to use the find.Overlapping.mzs function below to optimize the retention time tolerance threshold. In step four, a paired t-test & a Spearman correlation test is used to compare the intensity levels of the metabolites only for the redundant features that have m/z and retention time within defined tolerance levels as described above. Features with minimum median PID (or median CV; for more than two technical replicates) are chosen as representatives of each sub-group, and added to the final list. This scheme allows identification of unique features, and selection of the most consistent feature as a representative for features that overlap.

Value

Returns a matrix of columns of unique m/z values, elution times, signal strengths in each spectrum after merging results

Author(s)

Karan Uppal <[e-mail]>

yufree/xMSanalyzer documentation built on May 4, 2019, 6:35 p.m.