pleiofdr error

[rezarahman12]

Dear Developer Thank you for developing this software. I'm trying to run example datasets given pleiofdr. I'm getting an error regarding Manhattan plot generation stage. Error- ############################################################################### {Error using set Value must be a handle

Error in plot_Manhattan (line 219) set(get(gca,'XAxis'),'TickLength',[0.007 0.007])

Error in pleiotropy_analysis (line 272) h_Manhattan = plot_Manhattan(results, traitname1, traitnames, chrnumvec, options); }

Below is my config file-

!! Do not edit !!

Copy this file("config_default.txt") to "config.txt" and make your modifications there

reference file, such as ref9545380_1kgPhase3eur_LDr2p1.mat, which contains variables

LDmat (binary sparse LD matrix), chrnumvec, posvec, mafvec, is_ambiguous, is_intergenic.

Feel free to specify full path to the file.

reffile=/scratch/project_mnt/S0077/pleiofdr/ref9545380_1kgPhase3eur_LDr2p1.mat

Set path to the folder with traits files (PGC_SCZ_2014.mat, etc)

It is OK to keep 'traitfolder' empty. In this case 'traitfiles' must contain full path to each file.

Otherwise 'traitfiles' must give a file name within 'traitfolder', including .mat extension.

traitfolder=/scratch/project_mnt/S0077/pleiofdr/traitfolder

Set your GWAS file here for trait1 and other traits to condition on

traitfile1=CTG_COG_2018.mat traitname1=COG traitfiles={'SSGAC_EDU_2016.mat'} traitnames={'EDU'}

Output directory

outputdir=/scratch/project_mnt/S0077/pleiofdr/results

stattype can be either conjfdr (conjunctive fdr) or condfdr (conditional fdr).

recommended values of fdrthresh is 0.05 for conjfdr, and 0.01 for condfdr.

stattype=conjfdr fdrthresh=0.01

Random pruning options (on/off, number of iterations)

randprune=true randprune_n=500

Exclusion regions in format [CHR BP_from BP_to],

where CHR is chromosome and BP_from and BP_to are genomic coordinates (hg19 build).

To specify multiple regions use ;

Example for MHC and chr8 inversion: [6 25119106 33854733; 8 7200000 12500000]

exclude_chr_pos=[6 25119106 33854733]

Customizations for manhattan plot

manh_fontsize_genenames=12 manh_yspace=0.75 manh_ymargin=0.25 manh_colorlist=[1 0 0; 1 0.5 0 ; 0 0.75 0.75; 0 0.5 0; 0.75 0 0.75; 0 0 1; 0 1 0; 0 1 1]

An optional file containing additional reference information (e.g. the 9545380.ref file)

'SNP' column - for SNP rs# or other marker names

'A1' and 'A2' columns - alleles information

refinfo=

========= Other miscellaneous options (expert usage only) =========

Force random prune indices (pruneidx) to be re-generated for each run.

Setting this to false will preserve pruneidx across re-runs (unless you restart matlab or clear workspace).

Be careful with setting this to false.

Remember to clear matlab workspace whenever you change ldmatfile or randprune_n.

reset_pruneidx=true

randprune_repeats allows to choose from three resampling options ('default', 'maxout', 'none')

randprune_repeats=default

Threshold on Fisher's combined statistics (flp)

SNPs with flp below this thresh will not make it into the resulting loci table.

Set this parameter to 1 results to disable filtering on flp.

pthresh=1

Genomic correction options

perform_gc turns genomic correction on or off

use_standard_gc=true uses median genomic correction (standard from the literature),

while use_standard_gc=false imply our in-house developed genomic correction procedure (typically more conservative).

randprune_gc=true imply that lambdaGC factor will be calculated after random pruning.

perform_gc=true use_standard_gc=false randprune_gc=true

By default the exclusion list (exclude_chr_pos) is applied only when cond/conj FDR model is fit, but not during discovery.

Setting exclude_from_discovery=true will apply exclusion list also to the discovery.

This flag has not effect on 'exclude_ambiguous_snps' option.

exclude_from_discovery=false

Minor allele frequency threshold to exclude rare SNPs.

Note that if some SNPs have undefined (unknown) MAF they will be excluded too.

The likely reason for a SNP to have an undefined MAF is because it was not found in 1kG genotypes.

For such SNPs we also don't know their LD structure,

and hence can't perform random pruning on them.

You may set mafthresh=nan to keep such SNPs in the analysis,

but this is not recommended.

mafthresh = 0.005

A flag indicating whether to exclude ambiguous SNPs (A/T and C/G) from the analysis.

When set to 'true' ambigous SNPs will be excluded both from fit and from discovery,

regardless of the 'exclude_from_discovery' setting.

exclude_ambiguous_snps = true

A flag indicating whether to show plots on screen.

This applies to manhattan plot, QQ plots, Fold enrichment plots, and FDR lookup heatmap.

onscreen = true

pleioFDR analysis does not require effect dirrection, and is based on p-values.

For historical reason the pleioFDR pipeline require also zscore as input, and report files with z-score, to allow user to asses effect direction.

However, if effect direction info is not available for your GWAS summary statistics, then you may want to overwrite the above behavior.

Setting dummy_zscore = true will force pleioFDR code to calcualte z-score from p-value (forcing them to be positive).

dummy_zscore = false

A flag indicating whether to exit matlab upon completion of pleioFDR analysis.

exit_matlab_upon_completion = false ################ Could you please help me to avoid the error?

Kind regards Reza

[ofrei]

Hi Not sure what's wrong, perhaps you are using another version of Matlab that what we tested pleiofrd with. Could you try comment out the line where this fails? The line sets some fonts for Manhattan plot so it's optional. Or, comment out the whole mantlhattan plot section and make them using some other tool, e.g. qqman in R

[rezarahman12]

Dear @ofrei Thank you. Yes, I obtained the output except manhattan plot. So, according to your advice I comment out Manhattan plot section in plot_Manhattan.m file. It ran without any error. Thanks for your quick response. I would like to ask for another help on converting the result.mat file into csv. I tried to follow your instruction but for me it is a bit difficult to implement because I am not good at programming languages. Could you please give me a script that you used to convert the result.mat file into csv. I prepared 1 but need to be edited because I did by some google search. Could you please help me on the below to get desired results. I do highly appreciate your time and help. Best regards Reza

****code**** from scipy.io import loadmat data=loadmat("/scratch/project_mnt/S0077/pleiofdr/results/result.mat") x=data["mafvec"] print(x) print(type(x))

import pandas as pd df = pd.DataFrame(x)

print('DataFrame:\n', df)

default CSV

csv_data = df.to_csv() print('\nCSV String:\n', csv_data)