fixes

.tests/config/config.yaml

@@ -0,0 +1 @@
+../../config/config.yaml

config/config.yaml

@@ -0,0 +1,47 @@
+reference:
+  species: homo_sapiens
+  build: GRCh38
+  # ensembl release
+  release: 100
+
+samples:
+  rna:
+    # either specify sra: <accession>, or fq1: <path> and fq2: <path>
+    sra: SRR8758323
+    # denote read geometry as expected by salmon alevin
+    geometry:
+      reads: "2[1-end]"
+      barcodes: "1[1-16]"
+      umis: "1[17-26]"
+  adt:
+    # either specify sra: <accession>, or fq1: <path> and fq2: <path>
+    sra: SRR8758325
+    # denote read geometry as expected by salmon alevin
+    geometry:
+      reads: "2[1-15]"
+      barcodes: "1[1-16]"
+      umis: "1[17-26]"
+  hto:
+    # either specify sra: <accession>, or fq1: <path> and fq2: <path>
+    sra: SRR8758327
+    # denote read geometry as expected by salmon alevin
+    geometry:
+      reads: "2[1-15]"
+      barcodes: "1[1-16]"
+      umis: "1[17-26]"
+
+antibodies:
+  adt-seqs: data/adt.tsv
+  hto-seqs: data/hto.tsv
+
+thresholds:
+  # After the first execution of the workflow,
+  # determine an upper bound for HTO counts to consider
+  # by looking at results/plots/hto-counts.pdf,
+  # see https://combine-lab.github.io/alevin-fry-tutorials/2021/af-feature-bc.
+  max-hto-count: null
+  # After the second execution of the workflow,
+  # determine an upper bound for ADT counts to consider
+  # by looking at results/plots/adt-counts.pdf,
+  # see https://combine-lab.github.io/alevin-fry-tutorials/2021/af-feature-bc.
+  max-adt-count: null

workflow/Snakefile

@@ -1,3 +1,5 @@
+configfile: "config/config.yaml"
+
 include: "rules/common.smk"
 include: "rules/utils.smk"
 include: "rules/ref.smk"

workflow/rules/analysis.smk

@@ -5,9 +5,9 @@ rule seurat:
         hto="results/alevin-fry/quant/hto",
         geneid2name="resources/reference/geneid2name.tsv",
     output:
-        "results/seurat/{sample}.all.rds",
+        "results/seurat/all.rds",
     log:
-        "logs/seurat/{sample}.log",
+        "logs/seurat.log",
     conda:
         "../envs/seurat.yaml"
     script:

workflow/rules/common.smk

@@ -9,21 +9,19 @@ def get_reference(wildcards):
 
 def get_reads(wildcards):
     sample = config["samples"][wildcards.sample]
+    print(sample)
     if "fq1" in sample and "fq2" in sample:
         return {"fq1": sample["fq1"], "fq2": sample["fq2"]}
     else:
+        accession = sample["sra"]
         return {
-            "fq1": f"resources/sra/{sample[sra]}_1.fastq",
-            "fq2": f"resources/sra/{sample[sra]}_2.fastq",
+            "fq1": f"resources/sra/{accession}_1.fastq",
+            "fq2": f"resources/sra/{accession}_2.fastq",
         }
 
 
-def get_salmon_index(wildcards):
-    return f"resources/salmon-index/{get_unit_type(wildcards)}"
-
-
-def get_unit_type(wildcards):
-    return get_unit(wildcards)["type"]
+def get_sample(wildcards):
+    return config["samples"][wildcards.sample]
 
 
 def get_targets():

workflow/rules/read_mapping.smk

@@ -4,7 +4,7 @@ rule salmon_index:
     output:
         directory("resources/salmon-index/{sample}"),
     log:
-        "logs/salmon-index/{ref}.log",
+        "logs/salmon-index/{sample}.log",
     threads: 2
     params:
         extra="--features -k7",
@@ -15,20 +15,20 @@ rule salmon_index:
 rule salmon_alevin:
     input:
         unpack(get_reads),
-        index=get_salmon_index,
+        index="resources/salmon-index/{sample}",
     output:
         directory("results/salmon-alevin/{sample}"),
     params:
-        unit=get_unit,
+        sample=get_sample,
     log:
-        "logs/salmon-alevin/{sample}-{type}.log",
+        "logs/salmon-alevin/{sample}.log",
     conda:
         "../envs/salmon.yaml"
     shell:
         "salmon alevin "
         "-l ISR -i {input.index} -1 {input.fq1} -2 {input.fq2} "
-        "--read-geometry {params.unit[read_geometry]} --bc-geometry {params.unit[barcode_geometry]} "
-        "--umi-geometry {params.unit[umi_geometry]} -o {output} --sketch -p 16 "
+        "--read-geometry {params.sample[geometry][reads]} --bc-geometry {params.sample[geometry][barcodes]} "
+        "--umi-geometry {params.sample[geometry][umis]} -o {output} --sketch -p 16 "
         "2> {log}"
 
 

workflow/rules/ref.smk

@@ -59,11 +59,11 @@ rule spoof_t2g:
     input:
         get_reference,
     output:
-        "resources/reference/{type}/t2g.tsv",
+        "resources/reference/{sample}/t2g.tsv",
     log:
-        "logs/spoof-t2g/{type}.log",
+        "logs/spoof-t2g/{sample}.log",
     wildcard_constraints:
-        type="adt|hto",
+        sample="adt|hto",
     conda:
         "../envs/unix.yaml"
     shell: