Running MTAG without SNPIDs (SNP column having chr:pos).

[giuliaatosetto-netizen]

Hi,

Thank you for developing this software, I found it very useful.

I am wondering if it would be possible for MTAG to run with a SNP column containing chr:pos instead of a proper SNPID. I have tried to run it and found that both files I want to analyse read properly, but then I get this line in the .log file "2025/08/01/04:20:08 PM After merging with reference panel LD, 0 SNPs remain". This means there's no further analysis as there was no match to the reference LD panel. Does the LD panel just accept SNPIDs to do the match?

Thank you in advance.

Kind regards,
Giulia Tosetto

[paturley]

Hi Giulia, I believe that MTAG should be able to accept chr:pos format, but whatever format you use needs to match the format of your LD reference panel. Are you using the same format for both? Patrick

…

On Mon, Aug 4, 2025 at 5:20 AM giuliaatosetto-netizen < ***@***.***> wrote: *giuliaatosetto-netizen* created an issue (JonJala/mtag#241) <#241> Hi, Thank you for developing this software, I found it very useful. I am wondering if it would be possible for MTAG to run with a SNP column containing chr:pos instead of a proper SNPID. I have tried to run it and found that both files I want to analyse read properly, but then I get this line in the .log file "2025/08/01/04:20:08 PM After merging with reference panel LD, 0 SNPs remain". This means there's no further analysis as there was no match to the reference LD panel. Does the LD panel just accept SNPIDs to do the match? Thank you in advance. Kind regards, Giulia Tosetto — Reply to this email directly, view it on GitHub <#241>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFBUB5MK7PF5MFOYXTM2EN33L4QXDAVCNFSM6AAAAACDBOR4MWVHI2DSMVQWIX3LMV43ASLTON2WKOZTGI4DQNRXGE3DAOA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[okoropaulc]

Hi Giulia,

I believe that MTAG should be able to accept chr:pos format, but whatever
format you use needs to match the format of your LD reference panel. Are
you using the same format for both?

Patrick
…

Hi @paturley

Please I want to join my issue to the one raised here. I had the same error as the user above. When the snpid is chr:pos, mtag throws the same error of "After merging with reference panel LD, 0 SNPs remain". But when I change the snpid from chr:pos to rsid, it works and passes this stage. Even though I passed this step, I encountered another error ; KeyError: 'CAGCCAGC'

My questions are;

1. What LD reference panel does mtag use by default?
2. How can I pass my own LD reference panel to mtag when running mtag.py?
3. Could you advise on how I can resolve the key error issue. See below for the full list of error message;

2025/08/11/11:06:27 AM Trait 2: Dropped 34259 SNPs for duplicate values in the "snp_name" column
2025/08/11/11:06:32 AM Dropped 1740408 SNPs due to strand ambiguity, 9574547 SNPs remain in intersection after merging >
2025/08/11/11:06:47 AM Dropped 48081 SNPs due to inconsistent allele pairs from phenotype 2. 9519904 SNPs remain.
2025/08/11/11:06:52 AM Dropped 0 SNPs due to strand ambiguity, 9519904 SNPs remain in intersection after merging trait2
2025/08/11/11:06:52 AM ... Merge of GWAS summary statistics complete. Number of SNPs: 9519904
2025/08/11/11:07:07 AM Using 9519904 SNPs to estimate Omega (0 SNPs excluded due to strand ambiguity)
2025/08/11/11:07:07 AM Estimating sigma..
2025/08/11/11:07:29 AM 'CAGCCAGC'
Traceback (most recent call last):
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/mtag.py", line 1577, in
mtag(args)
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/mtag.py", line 1358, in mtag
args.sigma_hat = estimate_sigma(DATA[not_SA], args)
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/mtag.py", line 472, in estimate_sigma
rg_results = sumstats_sig.estimate_rg(args_ldsc_rg, Logger_to_Logging())
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/ldsc_mod/ldscore/sumstats.py", line 442, in estimate_rg
loop = _read_other_sumstats(args, log, None, sumstats, ref_ld_cnames,sumstats2=p2)
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/ldsc_mod/ldscore/sumstats.py", line 494, in read_other>
loop['Z2'] = _align_alleles(loop.Z2, alleles)
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/ldsc_mod/ldscore/sumstats.py", line 567, in _align_allel>
z *= (-1) ** alleles.apply(lambda y: FLIP_ALLELES[y])
File "/home/naj854/anaconda3/envs/my_py_2_7/lib/python2.7/site-packages/pandas/core/series.py", line 3591, in apply
mapped = lib.map_infer(values, f, convert=convert_dtype)
File "pandas/_libs/lib.pyx", line 2217, in pandas._libs.lib.map_infer
File "/n/data2/hsl/medicine/yau/lab/ANALYSES/jared/mtag/mtag/ldsc_mod/ldscore/sumstats.py", line 567, in
z *= (-1) ** alleles.apply(lambda y: FLIP_ALLELES[y])
KeyError: 'CAGCCAGC'
2025/08/11/11:07:29 AM Analysis terminated from error at Mon Aug 11 11:07:29 2025
2025/08/11/11:07:29 AM Total time elapsed: 4.0m:15.17s

[paturley]

Hi! (Feel free to correct me if I'm wrong about some of these software issues @JonJala )

1. MTAG is built on top of LD score regression, so I believe by default, MTAG just uses the LD scores from the software release correspond to when MTAG was released. I'm pretty sure that this is the HapMap3 SNPs of the 1kG Project EUR subsample, but it's been a while since I looked at this.
2. If you use the "--ld_ref_panel" flag, you should be able to pass in your own reference panel. Hopefully the help files has all the details you need to implement this.
3. MTAG has a few QC filters that only work for SNPs and not other types of variants. My guess is that your summary statistics include genetic variants where the effect or alternate allele is not A, T, C, or G but is instead CAGCCAGC, which is what is causing the error. If you drop any non-SNP variants from your data, I bet this error would go away.

Best,
Patrick

[giuliaatosetto-netizen]

Hi Patrick, thank you for your answer. I was using MTAG standard reference panel for my analysis, as I was not providing my own. When I change the chr:pos to SNPIDs, I am able to run the analysis, as the other user reports.

You mentioned in your reply to the other user that you are pretty sure the reference panel is HapMap 3 SNPs. How can I double-check which reference panel MTAG is using?

Kind regards,
Giulia

[paturley]

I'm pretty sure it's either in the documentation of the software or in the MTAG paper.

…

On Tue, Aug 12, 2025, 10:42 AM giuliaatosetto-netizen < ***@***.***> wrote: *giuliaatosetto-netizen* left a comment (JonJala/mtag#241) <#241 (comment)> Hi Patrick, thank you for your answer. I was using MTAG standard reference panel for my analysis, as I was not providing my own. When I change the chr:pos to SNPIDs, I am able to run the analysis, as the other user reports. You mentioned in your reply to the other user that you are pretty sure the reference panel is HapMap 3 SNPs. How can I double-check which reference panel MTAG is using? Kind regards, Giulia — Reply to this email directly, view it on GitHub <#241 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFBUB5KHCRCBJT3SVZ6BTY33NH4NXAVCNFSM6AAAAACDBOR4MWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTCNZZGY2TGMJQGA> . You are receiving this because you were mentioned.Message ID: ***@***.***>