--- author: Sam White toc-title: Contents toc-depth: 5 toc-location: left date: 2018-03-08 16:30:12+00:00 layout: post slug: assembly-geoduck-novaseq-using-sparseassembler-failed title: Assembly - Geoduck NovaSeq using SparseAssembler (failed) categories: - 2018 - Geoduck Genome Sequencing tags: - assembly - geoduck - NovaSeq - Panopea generosa - SparseAssembler --- Steven came across a [2012 paper in BMC Bioinformatics ("Exploiting sparseness in _de novo_ genome assembly")(https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-13-S6-S1) that utilized an assembly program we hadn't previously encountered: [SparseAssembler](https://sourceforge.net/projects/sparseassembler/) This software is intended to greatly reduce the required amount of RAM necessary to process very large assembly data sets. [As I previously learned, RAM is a limiting factor for assembly programs](https://robertslab.github.io/sams-notebook/posts/2018/2018-02-19-assembly-geoduck-illumina-novaseq-soapdenovo2-on-mox-fail/), and the install (if you can even call it that) was simply upacking a zip file ([program installations on Mox are not trivial](https://robertslab.github.io/sams-notebook/2018/02/21/novaseq-assembly-the-struggle-is-real-real-annoying/)so this seems like it has promise! The job was run on our Mox node. Here's the batch script to initiate the job: [20180308_soap_novaseq_geoduck_slurm.sh](https://owl.fish.washington.edu/Athaliana/20180308_SparseAssembler_novaseq_geoduck/20180308_soap_novaseq_geoduck_slurm.sh) [code lang=text] #!/bin/bash ## Job Name #SBATCH --job-name=20180308_sparse_assembler_geo_novaseq ## Allocation Definition #SBATCH --account=srlab #SBATCH --partition=srlab ## Resources ## Nodes (We only get 1, so this is fixed) #SBATCH --nodes=1 ## Walltime (days-hours:minutes:seconds format) #SBATCH --time=30-00:00:00 ## Memory per node #SBATCH --mem=500G ##turn on e-mail notification #SBATCH --mail-type=ALL #SBATCH --mail-user=samwhite@uw.edu ## Specify the working directory for this job #SBATCH --workdir=/gscratch/scrubbed/samwhite/20180308_SparseAssembler_novaseq_geoduck /gscratch/srlab/programs/SparseAssembler/SparseAssembler LD 0 NodeCovTh 1 EdgeCovTh 0 k 117 g 15 PathCovTh 100 GS 2200000000 i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/AD002_S9_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/AD002_S9_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/AD002_S9_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/AD002_S9_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR005_S4_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR005_S4_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR005_S4_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR005_S4_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR006_S3_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR006_S3_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR006_S3_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR006_S3_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR012_S1_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR012_S1_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR012_S1_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR012_S1_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR013_AD013_S2_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR013_AD013_S2_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR013_AD013_S2_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR013_AD013_S2_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR014_AD014_S5_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR014_AD014_S5_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR014_AD014_S5_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR014_AD014_S5_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR015_AD015_S6_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR015_AD015_S6_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR015_AD015_S6_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR015_AD015_S6_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR019_S7_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR019_S7_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR019_S7_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR019_S7_L002_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR021_S8_L001_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR021_S8_L001_R2_001_val_2_val_2.fastq i1 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR021_S8_L002_R1_001_val_1_val_1.fastq i2 /gscratch/scrubbed/samwhite/20180129_trimmed_again/NR021_S8_L002_R2_001_val_2_val_2.fastq [/code] ##### Results Output folder: [20180308_SparseAssembler_novaseq_geoduck/](https://owl.fish.washington.edu/Athaliana/20180308_SparseAssembler_novaseq_geoduck/) Well, this failed, but not because of memory issues (which is a good start)! Instead, it failed because the kmer size was too large??!! See the slurm output log file: * [slurm-142382.out](https://owl.fish.washington.edu/Athaliana/20180308_SparseAssembler_novaseq_geoduck/slurm-142382.out) [Kmergenie had indicated a kmer size of 117bp.](https://robertslab.github.io/sams-notebook/posts/2018/2018-02-05-novaseq-assembly-trimmed-geoduck-novaseq-with-meraculous/) Will reduce kmer size and try again. Fingers crossed...