A problem when ruuning runGAIA function in gaia package when analysis CNV data from TCGA
0
0
Entering edit mode
11 days ago
Dude • 0

Hi everyone! Sorry to bother, I encountered an irritating problem when running runGAIA function on CNV data from TCGA. The runGAIA function code is listed below

> system.time(suppressWarnings({
+   results <- runGAIA(cnv_obj,
markers_obj,
output_file_name = paste0("/Users/A/B/C/D/E/F/G/H/GAIA_", "LIHC", "_flt.txt"),
aberrations = -1,
chromosomes = -1,
num_iterations = 10,
threshold = 0.25)


}))

It takes more than an hour at least to run the code above each time, and I just keep getting the error message below, which is really frustrating.

Running Homogeneous peel-off Algorithm With Significance Threshold of 0.25 and
Homogenous Threshold of 0.12
.Error in if (tmp_list[[1]] == -1) { : the condition has length > 1
Timing stopped at: 4098 159.9 4368


I would really appreciate my question can be answered, thank you in advance!!!

# --------------------------------------------------

Ihe data i used can be found in the link below:

https://github.com/Datapioneer/Question.git

TCGA CNV GAIA • 223 views
1
Entering edit mode

Things to check include:

• are overlapping CN segments permitted in your input data?
• 'chr' prefix versus no 'chr' prefix
• which contigs are you processing? - -1 will mean that everything is processed, but what if you have weird contigs / patch sequences in your input?
• significance threshold may be too low and needs to be set to 1 to permit everything through on a first pass

I had a GAIA workflow adapted from Tiago Silva split across multiple answers, starting HERE ( How to extract the list of genes from TCGA CNV data ), if you wish to have a running example to follow as guidance.

0
Entering edit mode

Thank you, I've read the instructions you provided.

There is something that confuses me still a bit:

1. Is "overlapping CN segments" referring to the overlapped chromosome CNV segment in the same sample or between samples?
2. Does 'chr' prefix means that I need to convert my input cnvMatrix(CNVdata) data from data.frame into tibble and turn the "Chromosome" variable in both cnvMatrix and markersMatrix_filtered data from "<int> " to " "<chr>" ? Cause I just noticed that my cnvmatrix data does not contain variable attribution information between variable name row and the first observation row when using the command "head(CNVdata)" below:

Sample.Name Chromosome Start End Num.of.Markers Aberration

1 TCGA-DD-AAEH-01A-11D-A40Q-01 1 3218610 9454865 3653 0
2 TCGA-DD-AAEH-01A-11D-A40Q-01 1 16751233 16753901 3 0
3 TCGA-DD-AAEH-01A-11D-A40Q-01 1 111014215 111014420 3 0
4 TCGA-DD-AAEH-01A-11D-A40Q-01 1 149879545 151598232 615 1
5 TCGA-DD-AAEH-01A-11D-A40Q-01 1 151600441 151600454 2 0
6 TCGA-DD-AAEH-01A-11D-A40Q-01 1 151602647 247813706 60724 1

And below is my markersmatrix_filtered, whose "Chromosome" is presented as <int>, which probably mean that it should be changed to <chr>?

>   head(markersMatrix_filtered)
> #  A tibble: 6 × 3
>   Probe.Name Chromosome Start
>    <chr>           <int> <dbl>
> 1 CN_473964           1 61808
>  2 CN_473965           1 61823
>  3 CN_477984           1 62152
>  4 CN_473981           1 62920
>  5 CN_473982           1 62937
>  6 CN_497980           1 72704

1. Where and how can I check and specify the "contigs" parameter you just mentioned?

Sorry, I am still learning, hope I am not bothering, really appreciate it!!