Release v0.9.1

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Changed

• updated snakemake
  • effective genome size is now estimated per kmer length instead of per sample since checkpoints should work again.

Release v0.9.0

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Changed

• renamed most globals in uppercase (main exceptions are config and samples, treps and breps)
• moved most configuration steps into functions (reducing the number of stray globals)
• replaced static functions with dictionaries
• moved replicate stuff to the configuration
• Updated Salmon
• Added the option for Salmon to use the full genome as decoy sequence
• Salmon now uses the full genome as decoy sequence by default.
  • Config option quantifier_decoys controls which level of decoy aware quantification you want (options are 'none', 'partial' and 'full')
  • Option 'partial' is insanely memory intensive, and the Salmon docs suggest no benefit...
• improved parsing of the samples.tsv. More errors early on, to prevent headache later!

Fixed

• get_fastq_pair_reads() was using one sample, not any sample
• error message not working when trimming in scRNA-seq
• trackhubs when using a mix of stranded and unstranded datasets
• fix samples.tsv checks for forbidden symbols

Release v0.8.0

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Added

• idr call is configurable (idr_options)
• single-cell DESeq2 (currently only via deseq2science with user-specified groups per cell)
• scRNA quality control workflow with singleCellTK
  • cell calling/filtering with DropletUtils
  • mitochondrial gene set detection/filtering
  • doublet identification/filtering with scDblFinder
  • processing of alternative experiments, such as spike-in expression
  • qc report generation for cell/droplet based experiments
• added Seurat and FlatFile format export to scRNA qc workflow
• added parameter to select velocity matrix for qc and export

Changed

• raw/processed scRNA count tables are now stored and exported to SingleCellExperiment S4 objects instead of Seurat S4 objects
• moved scRNA post processing to separate module
• export unspliced velocity counts to separate sce object
• seq2science should be less susceptible to poor programming environment management by using the conda-ecosystem-user-package-isolation package
• seq2science will now demand all requirements exactly the way it likes it
  • this will make the workflows more stable.
• local fastq files are no longer renamed (and should just work)
• scRNA-seq trimming code simplified

Removed

• removed scRNA merging rule due to memory issues with large and sparse samples
• removed deprecated scRNA post-processing workflow (superseded by singleCellTK qc workflow)

Fixed

• fixed bug causing incorrect genome string in read_kb_counts.R
• bams generated with(out) filtering on size and tn5 shifting weren't removed when not necessary anymore

Changed

• rna-seq creates a TPM table for each quantification method

Release v0.7.2

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Added

• TPM to gene counts conversion with pytxi
  • by default attempts to use the GTF file to convert transcript_ids to gene_names
  • otherwise will use MyGene.info
• config option tpm2counts to chose which TPM to counts converter to use

Changed

• pytxi is now the default TPM to gene counts converter (over tximeta)
• peak/gene counts tables now use descriptive names (if given)
• MultiQC DESeq2 correlation plots now display correlation metrics in the figure
• using awful practices to eliminate checkpoint strandedness
• deeptools_flags renamed to deeptools_bamcoverage
• rna-seq trackhub per base tracks by default instead of bins per 50

Fixed

• edge-cases where seq2science was too strict with rerunning
• assembly stats log scale on the y-axis
• s2s explain wont tell you about subsampling to -1 (all) reads
• tn5 shift cigar string parsing edge-case (reference deletions/insertions)

Release v0.7.1

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Fixed

• issue with broad peaks and upsetplots

Release v0.7.0

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Biggest change is that we revert back to snakemake 5.18 since higher versioned snakemake's cause instability.

Added

• upset plot as QC for peak calling. Should give a first feeling about the distribution of peaks between samples/conditions.

Changed

• downgraded the snakemake backend as snakemake 6+ is unstable for us.

Fixed

• corrupt environment creation with libreadline for edgeR normalization.
• subsampling causing a crash caused by bad syntax.

Release v0.6.1

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Fixed

• corrupt environment creation with libcrypto in combination with strandedness rule

Release v0.6.0

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and (sc)RNA-seq workflows.

Release 0.6.0 is a mix of bug fixes, small changes, and bigger stuff. Most importantly:

• added a deseq2science command to do differential expression analysis on user-supplied tables with seq2science settings
• for single-cell RNA-seq ADT-quantification is possible
• snakemake library updated, giving seq2science a new-ish look :)

The full changes are listed below:

Added

• added generic stats to the MultiQC report about the assembly, which might help pin point problems with the assembly used.
• added the slop parameter to the config.yaml of atac-seq and chip-seq workflows, just so they are more visible.
• added support for seurat object export and merging for kb workflow.
• added support for CITE-seq-count for ADT quantification
• added the option to downsample to a specific number of reads.
• new deseq2science command

Changed

• Seq2science now makes a separate blacklist file per blacklist option (encode & mitochondria), so that e.g. RNA-seq and ATAC-seq workflows can be run in parallel and don't conflict on the blacklist.
• error messages don't show the full traceback anymore, making it (hopefully) more clear what is going wrong.
• The effective genome size is now not calculated per sample, but per read length. When dealing with multiple samples (of similar) length this improves computational burden quite some.
• samtools environment updated to version 1.14

Fixed

• config option slop is now passed along to each rule using it
• edge-case where local samples are in the cache, but not present in the fastq_dir
• bug with differential peak/gene expression across multiple assemblies
• bug with kb ref not creating index for non-velocity analysis
• bug with count import in read_kb_counts.R for technical replicates and meta-data handling
• deseq2 ordering in multiqc report
• issue with slop not being used for the final count table
• bug with onehot peaks not reporting the sample names as columns

Release v0.5.6

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and RNA-seq workflows.

Added

• MA plot, volcano plot, and PCA plots added to QC report for deseq2 related workflows

Changed

• updated salmon & tximeta versions
• colors for DESeq2 distance plots "fixed"
• updated bwa-mem2 version and reduced the expected memory usage of bwa-mem2 to 40GB
• seq2science now uses snakemake-minimal

Fixed

• stranded bigwigs are no longer inverted (forward containing reverse reads and vice-versa).
• fix in rename_sample preventing the inversion of R1 and R2 FASTQs.
• bug with parsing cli for explanations
• show/hide buttons for treps are actually made for multiqc report
• fixes in deseq2/utils.R
  • the samples.tsv will now work with only 2 columns
  • the samples.tsv column names will be stripped of excess whitespace, similar to the config.
• ATAC-seq pipeline removing the final bams, keeping the unsorted one

Release v0.5.5

Automated preprocessing of Next-Generation Sequencing data, including full (sc)ATAC-seq, ChIP-seq, and RNA-seq workflows.

Changed

• duplicate read marking happens before sieving and no reads get removed. Removal of duplicate reads now controlled with flag remove_dups in the config.
• changed option heatmap_deeptools_options to deeptools_heatmap_options
• Updated sra tools and parallel fastq-dump versions
• Updated genomepy version
• Gene annotations are no longer gzipped and ungzipped. This should reduce rerunning.

Fixed

• rerunning being triggered too easily by input order
• issue with qc plots and broad peaks
• magic with prefetch not having the same output location on all machines
• issue with explain having duplicate lines