SNPBag Inference Server
Please enter a valid email address to receive your result.
Please enter a valid email address
General Instructions:
- Four tasks are available: imputation, embedding, ancestry and relatedness inference.
- Recommended workflow: task 1 -> 2 -> 3 for ancestry inference.
- Recommended workflow: task 1 -> 2 -> 4 for relatedness inference.
- In each task, you can upload your own files or use demo files.
- You will see results right away or get them by email in minutes.
- Check spam if you do not get the email.
- If a task fails (e.g., 'Out of GPU memory' or 'CUDA-capable device(s) is/are busy'), try again.
- Feedbacks are welcome at https://github.com/augix/SNPBag/issues.
- Publications: bioRxiv.
- All individuals are from 1000 Genome Project.
- populations_omni.vcf.gz: 10 individuals from different populations.
- family_CHS_affy6.vcf.gz: a Chinese family of 6 individuals.
- family_PEL_ukb.vcf.gz: a Peruvian family of 6 individuals.
- omni: Omni SNP array.
- affy6: Affymetrix 6.0 SNP array.
- ukb: UK Biobank SNP array.
- Population and pedigree information from 1000 Genome Project: link.
Session ID: Loading...
Email:
Email:
Task 1: Genotype Imputation
- This fills in missing SNP genotypes.
- Upload a compressed VCF file (
.vcf.gz) including CHROM, ID (rsID), REF, ALT, and FORMAT columns. - ID must be rsID.
- See the demo file for the required format.
- Multiple individuals per file are supported. It takes 5 and 10 minutes for 32 individuals running on Nvidia A800 and RTX3090.
Analysis Result:
Download Result File
Demo data
Task 2: Genome Embedding
- This task creates a compressed embedding (representation) for each input genome.
- Upload an imputed VCF (.gz) file including CHROM, ID (rsID), REF, ALT, FORMAT columns.
- Result: a zip file with .emb binary files.
- Multiple individuals are supported. It takes ~20 min for 32 individuals on Nvidia A800.
Analysis Result:
Download Result File
Demo data
Task 3: Population Classification
- This does ancestry inference for individuals with genome embeddings.
- Upload an embedding file (.emb) generated from the Embedding Generation task (task 2) for each individual.
- The model includes 26 populations defined by the 1000 Genomes Project.
- Demo: HG00155 (GBR), HG00171 (FIN), HG01552 (CLM), HG01567 (PEL), HG01583 (PJL), HG02056 (KHV), HG02463 (GWD), HG03008 (BEB), NA18939 (JPT), NA19434 (LWK)
Demo data
Task 4: Relatedness Inference
- This infers the degree of relatedness between two individuals based on their genome embeddings.
- Upload two embedding files (.emb) generated by the Embedding Generation task (task 2) for each individual.
- Relatedness 0 means unrelated (or > 12th degree relative).
- Relatedness 1 means 1st degree relative (parent/child, full sibling).
- Relatedness 2 means 2nd degree relative (grandparent/grandchild, aunt/uncle/niece/nephew)
- More information on Wikipedia about Coefficient of relationship
- Demo:
