From c626bd043829cedc24eeb9f1166373550b06f038 Mon Sep 17 00:00:00 2001
From: Mervin Fansler <mmfansler@gmail.com>
Date: Fri, 2 Jul 2021 15:24:59 -0400
Subject: [PATCH] adds heart_10k example

---
 Snakefile                                    |  2 +-
 examples/heart_10k_v3_fastq/config.yaml      | 17 +++++++++++++++++
 examples/heart_10k_v3_fastq/download.sh      | 15 +++++++++++++++
 examples/heart_10k_v3_fastq/sample_sheet.csv |  2 ++
 examples/neuron_10k_v3_fastq/download.sh     |  2 +-
 5 files changed, 36 insertions(+), 2 deletions(-)
 create mode 100644 examples/heart_10k_v3_fastq/config.yaml
 create mode 100644 examples/heart_10k_v3_fastq/download.sh
 create mode 100644 examples/heart_10k_v3_fastq/sample_sheet.csv

diff --git a/Snakefile b/Snakefile
index cdcd7b0..90fd5e0 100644
--- a/Snakefile
+++ b/Snakefile
@@ -204,6 +204,6 @@ rule sce_txs_to_genes:
     output:
         sce="data/sce/{dataset}.genes.Rds"
     resources:
-        mem_mb=8000
+        mem_mb=16000
     script:
         "scripts/sce_txs_to_genes.R"
diff --git a/examples/heart_10k_v3_fastq/config.yaml b/examples/heart_10k_v3_fastq/config.yaml
new file mode 100644
index 0000000..e6fca27
--- /dev/null
+++ b/examples/heart_10k_v3_fastq/config.yaml
@@ -0,0 +1,17 @@
+sample_file:
+  "examples/heart_10k_v3_fastq/sample_sheet.csv"
+
+annotation_file:
+  "examples/heart_10k_v3_fastq/annots.csv"
+
+bx_whitelist:
+  "extdata/bxs/3M-february-2018.txt"
+
+final_output_file:
+  "data/sce/heart_10k_v3_fastq.utrome.txs.Rds"
+
+tech:
+  "10xv3"
+
+strand:
+  "--fr-stranded"
diff --git a/examples/heart_10k_v3_fastq/download.sh b/examples/heart_10k_v3_fastq/download.sh
new file mode 100644
index 0000000..13da67e
--- /dev/null
+++ b/examples/heart_10k_v3_fastq/download.sh
@@ -0,0 +1,15 @@
+#!/usr/bin/env bash
+
+wget -O tmp.tar https://cg.10xgenomics.com/samples/cell-exp/3.0.0/heart_10k_v3/heart_10k_v3_fastqs.tar \
+  && tar -xf tmp.tar \
+  && mv heart_10k_v3_fastqs fastq \
+  && rm tmp.tar
+
+## Download analysis data and make annots.csv
+wget -O tmp.tar.gz https://cf.10xgenomics.com/samples/cell-exp/3.0.0/heart_10k_v3/heart_10k_v3_analysis.tar.gz \
+    && tar -xvzf tmp.tar.gz \
+    && rm tmp.tar.gz
+
+join -t',' analysis/clustering/graphclust/clusters.csv analysis/tsne/2_components/projection.csv | \
+    sed -E 's/([ACGT]{16})-1/heart_10k_v3_fastq_\1/' | \
+    sed '1 s/^.*$/cell_id,cluster,tsne_1,tsne_2/' > annots.csv
diff --git a/examples/heart_10k_v3_fastq/sample_sheet.csv b/examples/heart_10k_v3_fastq/sample_sheet.csv
new file mode 100644
index 0000000..6a15d91
--- /dev/null
+++ b/examples/heart_10k_v3_fastq/sample_sheet.csv
@@ -0,0 +1,2 @@
+sample_id,file_type,files
+heart_10k_v3_fastq,fastq,examples/heart_10k_v3_fastq/fastq/heart_10k_v3_S1_L001_R1_001.fastq.gz;examples/heart_10k_v3_fastq/fastq/heart_10k_v3_S1_L001_R2_001.fastq.gz;examples/heart_10k_v3_fastq/fastq/heart_10k_v3_S1_L002_R1_001.fastq.gz;examples/heart_10k_v3_fastq/fastq/heart_10k_v3_S1_L002_R2_001.fastq.gz
diff --git a/examples/neuron_10k_v3_fastq/download.sh b/examples/neuron_10k_v3_fastq/download.sh
index 8cd9e56..88ee0ac 100644
--- a/examples/neuron_10k_v3_fastq/download.sh
+++ b/examples/neuron_10k_v3_fastq/download.sh
@@ -1,4 +1,4 @@
-#!/usr/bin/env #!/usr/bin/env bash
+#!/usr/bin/env bash
 
 wget -O tmp.tar https://cg.10xgenomics.com/samples/cell-exp/3.0.0/neuron_10k_v3/neuron_10k_v3_fastqs.tar \
   && tar -xf tmp.tar \