Still getting ERROR:001 grch38 column header issues...

[Samuel Marsh]

Hi,

Thank you so much for maintenance and update of this fantastic tool and really appreciate the update to allow selection of grch38 sum stats for snp2gene!

I'm running into issue and not sure why it's still failing.  I've read through the common issues and solutions and can't seem to make sense of this one.  Gone through many of previous ERROR:001 problems.

The relevant job IDs are: 748899 and 748904.

The input files are loaded into R and columns filtered to reduce file size and reordered to match required grch38 column instructions.

For job 748899 I had forgotten to set quote = F when using write.table in R.  When I set quote = F for next job it seemed to have issue recognizing the header name separation (my best guess from the log files).

After writing tzv file in R I'm just using gzip via terminal to compress file before uploading to run.

just normal command: gzip file_name.tsv

Is there something else that I'm missing during the file writing coming out of R that's causing issue?

Also just wanted to check (since I haven't been able to get passed column name recognition issues) but when running in grch38 mode is it allowed to have a 7th column specifying total N/per SNP to be specified in Sample (N) parameter section?  Or will that extra column cause error?

Thanks very much!!

Sam

Full R code being run:

```

library(tidyverse)
library(data.table)

bellenguez_prev <- fread("~/Dropbox/GWAS/GWAS Summary Stats/Bellenguez_filtered.tsv.gz", data.table = F)

bellenguez_reorder <- bellenguez_prev %>%
  select(chromosome, base_pair_location, effect_allele, other_allele, beta, p_value, n_total)

write.table(bellenguez_reorder, file = "~/Downloads/bellenguez_grch38_reordered.tsv", quote = FALSE)

```

[Tanya Phung]

Hi Samuel, 

For job 748899, this is the log: 

2026-06-26 16:31:22,169 [INFO] __main__: Input file header: ['chromosome', 'base_pair_location', 'effect_allele', 'other_allele', 'beta', 'p_value', 'n_total']
2026-06-26 16:31:22,169 [ERROR] __main__: When beta column is present, the header must be: chromosome, base_pair_location, effect_allele, other_allele, beta, p_value

The issue here is that there is an extra column n_total. At the current version of FUMA (v2.1.5), this is what you need to do: 

You need to remove the n_total column and use the effective sample size. 

Please see: https://fuma-docs.readthedocs.io/en/latest/snp2gene/prepare_input_files.html. For using GRCh38 option you need to provide exactly 3 columns or 6 columns. 

Your input file needs to be similar to this: https://github.com/tanyaphung/commonly_used_codes/blob/master/fuma_related/acceptable_snp2gene_sumstats/grch38/GCST90000255_GRCh38_chr21_fuma.tsv

You then put in the sample size in section 2 of the page: 

I understand that this is a currently a limitation and a update for this will be worked on soon, to accommodate for more flexibility in the input file. 

For job 748904, this is the log: 
2026-06-26 16:43:00,564 [INFO] __main__: Input file header: ['chromosome base_pair_location effect_allele other_allele beta p_value n_total']
2026-06-26 16:43:00,564 [ERROR] __main__: Columns chromosome and base_pair_location are required in the input file.

As you can see, it looks like the delimiter is incorrect. Therefore, FUMA could not parse the header file. 
Please check the link above for correct file format. 

Best,
Tanya

[Samuel Marsh]

Hi Tanya,

Thanks so much for quick reply!

That makes sense and now got things running smoothly.

On topic of total N.  Do you/team have recommendations for how to select effective sample size from sum stats that have per SNP N?

Also wondering if SNP2GENE would still function as previously if only rsID and stats values were provided and chromosome and base pair locations were omitted to use grch38?  Just wondering because that would allow for using file with N/SNP but still grch38.

Thanks again!!

Sam

[Tanya Phung]

Hi Sam, 

You can use resources from here for some information on computing effective sample size (https://isgw-forum.colorado.edu/t/backing-out-effective-sample-size/523) but please check with others as I am not an expert on this. 

Otherwise, you can also use total sample size for N if that is easier. I asked the developer of MAGMA, and this is what he said in terms of sample size: 

In general, sample size is used but it doesn't play a massive role. The primary uses are just 1) general reference/documentation, 2) when performing meta-analysis in MAGMA, and 3) accounting for possible differences in sample size between genes in the gene-set analysis. Obviously point 3 only applies if sample size per SNP is available, which is also why sample size per SNP is preferred if present.

For the most part, the difference between total and effective sample size doesn't matter overly much for this, since it's mainly about the relative differences. If it's available, I'd probably take the effective sample size, but if not then going with total sample size is fine.  

if only rsID and stats values were provided and chromosome and base pair locations were omitted to use grch38? 

In this case, you are correct. You can submit a file with rsID (removing chrom and position columns completely) then you can specify the N column. 

Best,
Tanya