ArrayPipe FAQ

Content:

1. Why does the normalization not work?
2. What do the different flag values mean?
3. What do the abbreviated data headers stand for?

1. Why does the normalization not work?

   There are several cases where the normalization procedure can fail or at least seems to fail:

   • You are trying to use printTipLoess and one or more of the sub-grids do not contain valid data, i.e. all the spots are flagged or undefined. This results in an error message like the following:
     

     Error in simpleLoess(y, x, w, span, degree, parametric, drop.square, normalize,  : 
            invalid `x'
     Execution halted
     

     Possible work-arounds for this are to either use limma loess (subgrid) or one of the loess normalization methods that work on the whole slide instead (see below). Please note that beside some technical differences, the default span is different between limma loess (subgrid) and marray's printTipLoess (0.3 and 0.4, respectively).
     
     
   • You are trying to use printTipLoess and one or more of the sub-grids contain only few data points, i.e. most of the spots are flagged or undefined. This results in an error message like the following:
     

     Error: span is too small
     Execution halted
     

     Go to the settings for this module and increase the span (0.4 by default). See external link for more information on span and loess.
     
     
   • You are trying to use Loess and one or more of the sub-grids do not contain valid data, i.e. all the spots are flagged or undefined. This results in an error message like the following:
     

     > maBoxplot(raw[,2], x = "maPrintTip", y = "maM", main = "12710739_Cy3.txt: pre--normalization")
     Error in boxplot.default(split(mf[[response]], mf[-response]), ...) : 
            names attribute [48] must be the same length as the vector [47]
     Execution halted
     

     In this case the normalization worked out but the boxplots which present the data before and after the transformation could not be created due to missing data. Please use the module 'Signal boxplot (print tip)' to generate these separately.
2. What do the different flag values mean?

   ArrayPipe uses integers to represent different flags. These numbers are chosen in a way that allows combining multiple flags into one number and breaking it up again into its components. The values used are as follows:

   Flag Name	Flag Value	Explanation
   no flag:	0	spot has no flag set
   automatic:	1	spot was flagged by quantification software
   markers:	2	set by 'Flag markers'
   dup-flaw:	4	set by 'Flag flawed duplicates'
   floor:	8	set by background correction and 'Set cutoffs'
   ceiling:	16	set by 'Set cutoffs'
   warning:	32	assigned to ImaGene spot values other than 1,2,5
   no_ratio:	64	set by functions attempting calculation of ratios
   error:	128	assigned to spots with invalid values
   undefined:	256	assigned to spots without values
   empty_dup:	512	set by 'Merge duplicate spots'
   low_quality:	1024	used for 'A' flags in TMEV files
   user1:	2048	used for 'U' flags in TMEV files
   list1:	4096	set by 'Filter by value'

   Any additional flags (list2, list3, or user-defined flags) will have a value that is double the amount of the previously highest one.

   Whereas most of the flags are assigned by ArrayPipe, the 'automatic' flag is read in from the data file and has normally been assigned automatically by the quantification software.

   Combining multiple flags happens through summing up their values, e.g. a flag of 34 represents a 'marker' (2) with a 'warning' (32).



3. What do the abbreviated data headers stand for?

   Rather than always writing out the words 'foreground intensities' and other descriptors of data types, ArrayPipe uses abbreviations to describe the data generated. The tables below explain the most common ones.

   Basic data types:

   ID	spot identifier
   FL	flag
   FG	foreground intensities
   BG	background intensities

   
   Special data types:

   FG_CRD	foreground intensities after background correction
   BG_CALC	calculated background values
   Norm	normalized values
   p-value	p-values from t-tests or permutations
   Z-score	Z-score from sliding window approach

   On top of that, ArrayPipe uses several appendices to further specify a data type.

   Appendices:

   _L	values are log-transferred
   _R	ratios
   _I	intensity product
   _M	merged data
   _i	number of items that were merged
   _SD	standard deviation

   These appendices can be combined, e.g., the term Norm_L_R stands for "ratios calculated from log-transferred, normalized data".

Home         Documentation  

---

last updated at Wed Jun 6 09:01:21 PDT 2007
for questions or remarks e-mail karsten_hokamp@sfu.ca.