Feature Request: Option to subset reference SNPs to query file positions before training

[weekend37]

Would enhance model performance in most cases and avoid disaster where imputation has not been performed.

[candevrivera2021]

Hi. I am having this issue when training the model from scratch after filtering the snps in the query file. I am not sure what to do.

After building the model, I get this error when doing the inference in the query samples. Any advice is appreciated.

Launching inference... Loading and processing query file...

• Number of SNPs from model: 333651
• Number of SNPs from file: 447132
• Number of intersecting SNPs: 333651
• Percentage of model SNPs covered by query file: 100.0%
• Found 24392 (7.3106%) different reference variants. Adjusting... Inferring ancestry on query data... /users/cvergara/gnomix/src/utils.py:313: PerformanceWarning: DataFrame is highly fragmented. This is usually the result of calling frame.insert many times, which has poor performance. Consider joining all columns at once using pd.concat(axis=1) instead. To get a de-fragmented frame, use newframe = frame.copy() df[data_samples[i]] = genotype_person