From 0ce141b41f858787ffcaf8930572a3c6b8671c4e Mon Sep 17 00:00:00 2001
From: khalid <khalid.belkhir@umontpellier.fr>
Date: Fri, 3 Dec 2021 16:34:12 +0100
Subject: [PATCH] Add snap index
---
tools/snap_index/snap_index.rule.snakefile | 25 ++++++++
tools/snap_index/snap_index.yaml | 66 ++++++++++++++++++++++
2 files changed, 91 insertions(+)
create mode 100755 tools/snap_index/snap_index.rule.snakefile
create mode 100755 tools/snap_index/snap_index.yaml
diff --git a/tools/snap_index/snap_index.rule.snakefile b/tools/snap_index/snap_index.rule.snakefile
new file mode 100755
index 00000000..fe8c9d8a
--- /dev/null
+++ b/tools/snap_index/snap_index.rule.snakefile
@@ -0,0 +1,25 @@
+rule <step_name>__snap_index:
+ input:
+ **<step_name>__snap_index_inputs()
+ output:
+ index = (
+ config["<step_name>__snap_index_output_dir"]+"/Genome",
+ config["<step_name>__snap_index_output_dir"]+"/GenomeIndex",
+ config["<step_name>__snap_index_output_dir"]+"/GenomeIndexHash",
+ config["<step_name>__snap_index_output_dir"]+"/OverflowTable",
+ ),
+ threads:
+ config["<step_name>__snap_index_threads"]
+ log:
+ config["results_dir"]+"/logs/" + config["<step_name>__snap_index_output_dir"] + "/index.log"
+ params:
+ command = config["<step_name>__snap_index_command"],
+ snap_index_seed_size = config["<step_name>__snap_index_seed_size"],
+ out_dir = config["<step_name>__snap_index_output_dir"]
+ shell:
+ "{params.command} "
+ "{input.genome_fasta} "
+ "{params.out_dir} "
+ "-s {params.snap_index_seed_size} "
+ "-t{threads} |& tee {log};" #no space after -t
+
diff --git a/tools/snap_index/snap_index.yaml b/tools/snap_index/snap_index.yaml
new file mode 100755
index 00000000..7c71ca81
--- /dev/null
+++ b/tools/snap_index/snap_index.yaml
@@ -0,0 +1,66 @@
+{
+ id: snap_index,
+ name: Scalable Nucleotide Alignment Program,
+ article: arXiv:1111.5572v1,
+ website: "https://www.microsoft.com/en-us/research/project/snap/",
+ git: "https://github.com/amplab/snap",
+ description: "fast and accurate aligner for short DNA reads. It is optimized for modern read lengths of 100 bases or higher, and align them quickly through a hash-based indexing scheme.",
+ version: "2.0",
+ documentation: "https://github.com/amplab/snap/tree/master/docs",
+ multiqc: "custom",
+ commands:
+ [
+ {
+ name: snap_index,
+ cname: "snap index",
+ command: snap-align index,
+ category: "indexing",
+ output_dir: snap/index,
+ inputs: [{ name: genome_fasta, type: "contigs", description: "Fasta reference", from: "parameter" }],
+ outputs: [{ name: index, type: "snap_index", file: index, description: "Index files for snap alignment" }],
+ options:
+ [
+ {
+ name: snap_index_threads,
+ prefix: -t,
+ type: numeric,
+ value: 4,
+ min: 1,
+ max: NA,
+ step: 1,
+ label: "Number of threads to use",
+ },
+ {
+ name: snap_index_genome_fasta,
+ type: input_file,
+ value: "",
+ label: "Path to reference genome fasta file",
+ },
+ {
+ name: snap_index_seed_size,
+ prefix: -s,
+ type: numeric,
+ value: 24,
+ min: 12,
+ max: 64,
+ step: 1,
+ label: "seed size",
+ }
+ ]
+ },
+ ],
+ install: {
+ snap: [
+ "cd /opt/biotools",
+ "git clone https://github.com/amplab/snap.git",
+ "cd snap",
+ "make -j 8",
+ "ENV PATH /opt/biotools/snap:$PATH",
+ ]
+ },
+ citations: {
+ snap: [
+ "Faster and More Accurate Sequence Alignment with SNAP. Matei Zaharia, William J. Bolosky, Kristal Curtis, Armando Fox, David Patterson, Scott Shenker, Ion Stoica, Richard M. Karp, and Taylor Sittler. arXiv:1111.5572v1, November 2011."
+ ]
+ }
+}
\ No newline at end of file
--
GitLab