How to create a sample sheet¶

The sample sheet is a CSV file describing your samples and their data locations. Each row represents one sequencing run for one sample.

For the full column specification, see the sample sheet schema reference.

Minimal example¶

Every sample sheet needs three columns: sample_id, input_type, and input. Here is a minimal sheet for two samples from the NCBI Sequence Read Archive:

sample_id,input_type,input
sample_A,srr,SRR12345678
sample_B,srr,SRR12345679

snpArcher will download the reads automatically.

From SRA accessions¶

To create a sample sheet from an existing NCBI BioProject:

1. Go to the BioProject page on the SRA.
2. In the Project Data table, click the link in the Number of Links column on the SRA Experiments row.
3. Click Send results to Run Selector near the top of the search results.
4. Select or deselect samples using the checkboxes.
5. Click the Metadata button in the Download column to download SraRunTable.txt.
6. Open the file and create a new CSV with three columns:

sample_id,input_type,input
SAMN12345678,srr,SRR12345678
SAMN12345679,srr,SRR12345679
SAMN12345680,srr,SRR12345680

Use the BioSample column from the SRA table as sample_id and the Run column as input. Set input_type to srr for every row.

From local FASTQ files¶

For local paired-end FASTQ files, set input_type to fastq and join the forward and reverse read paths with a semicolon:

sample_id,input_type,input
bird_1,fastq,/data/reads/bird_1_R1.fq.gz;/data/reads/bird_1_R2.fq.gz
bird_2,fastq,/data/reads/bird_2_R1.fq.gz;/data/reads/bird_2_R2.fq.gz
bird_3,fastq,/data/reads/bird_3_R1.fq.gz;/data/reads/bird_3_R2.fq.gz

From BAM files¶

To start from aligned BAM files (skipping trimming and alignment):

sample_id,input_type,input
sample_A,bam,/data/bams/sample_A.bam
sample_B,bam,/data/bams/sample_B.bam

Each sample must have exactly one BAM row.

From gVCF files¶

To start from per-sample gVCF files (skipping everything up to joint genotyping):

sample_id,input_type,input
sample_A,gvcf,/data/gvcfs/sample_A.g.vcf.gz
sample_B,gvcf,/data/gvcfs/sample_B.g.vcf.gz

Each sample must have exactly one gVCF row.

Mixing input types¶

You can mix input types in a single sample sheet. snpArcher will start each sample at the appropriate pipeline stage:

sample_id,input_type,input,library_id
sample_A,fastq,/data/sample_A_R1.fq.gz;/data/sample_A_R2.fq.gz,lib_A
sample_B,srr,SRR12345678,
sample_C,bam,/data/sample_C.bam,

Multiple runs per sample¶

When a sample has been sequenced across multiple lanes or runs, create one row per run with the same sample_id. snpArcher will align each run independently, then merge the BAMs before variant calling.

The library_id column controls duplicate-marking behavior during the merge:

• Rows with the same library_id are treated as technical replicates of one library (e.g., the same library sequenced on multiple lanes). Duplicates are marked across the merged set.
• Rows with different library_id values are treated as independent libraries. Duplicates are marked within each library separately.

sample_id,input_type,input,library_id
sample_A,fastq,/data/sample_A_lane1_R1.fq.gz;/data/sample_A_lane1_R2.fq.gz,lib1
sample_A,fastq,/data/sample_A_lane2_R1.fq.gz;/data/sample_A_lane2_R2.fq.gz,lib1
sample_A,fastq,/data/sample_A_newlib_R1.fq.gz;/data/sample_A_newlib_R2.fq.gz,lib2
sample_B,fastq,/data/sample_B_R1.fq.gz;/data/sample_B_R2.fq.gz,lib3

In this example, sample_A has three sequencing runs: two from the same library (lib1) and one from a different library (lib2).

Disabling duplicate marking¶

If duplicate marking is inappropriate for your protocol (e.g., amplicon sequencing), set mark_duplicates to false:

sample_id,input_type,input,mark_duplicates
sample_A,fastq,/data/sample_A_R1.fq.gz;/data/sample_A_R2.fq.gz,false
sample_B,fastq,/data/sample_B_R1.fq.gz;/data/sample_B_R2.fq.gz,false

You can also set the global default in config.yaml under reads.mark_duplicates and omit the column from the sample sheet.

Column reference¶

For the full specification and validation rules, see the sample sheet schema reference.