Use biocontainers images for star, gffread, salmon, kallisto, cutadapt, samtools, fastqc, alfa, bedtools, bedgraphtobigwig. Change container from bash to ubuntu. Fixes #149

BIOPZ-Iborra de Toledo Paula authored 4 years ago and BIOPZ-Gypas Foivos committed 4 years ago

The rules rely on https://github.com/zavolanlab/merge_kallisto
Update info in pipeline_documentation.md

BIOPZ-Katsantoni Maria authored 4 years ago and Alex Kanitz committed 4 years ago

fix: Renamed samples_concat.tsv to samples.multiple_lanes.tsv. Renamed rows with split with the same name as the other test samples, so that I do not change the tests (md5 and sunch). Removed the one lane samples. Created config that uses this tsv file

- clean up command line interface
  - improve descriptions
  - add consistent structure
  - remove or merge superfluous CLI arguments
  - set defaults
  - update test calls
  - update docs
  - when importing data from LabKey, table is saved to 'samples.tsv.labkey' in same directory as Snakemake sample table
- allow user to specify environment variables and relative paths in input table and on CLI
  - relative paths in the input table are interpreted with respect to the directory containing the input table
  - relative paths will are interpreted with respect to the current working directory; this is to achieve portability with respect to tests but is discouraged in production because its behavior is not very predictable from the user's perspective; consequently a warning is thrown
- set STAR index size to read length - 1
- remove `gtf_filtered` and `tr_fasta_filtered` and update Snakefiles and test sample tables accordingly
- rename some MultiQC report-related parameters and update Snakefiles and test config files accordingly
- add logging
- add docstrings to module and all functions
- add typing definitions to all functions
- restructure and comment code to improve readability
- linters `flake8` and `mypy` pass

BIOPZ-Katsantoni Maria authored 4 years ago and Alex Kanitz committed 4 years ago

* Sequencing mode-related changes:
  * allowed sequencing modes in Snakemake input table changed from `paired_end` and `single_end` to `pe` and `se`, respectively
  * remove sequencing mode from output paths for each rule
  * corresponding wild cards removed entirely from all rules that do not depend on sequencing mode (currently all rules that are defined in the main `Snakefile` in the project root directory)
  * where absolutely necessary, sequencing mode is added as part of output file or directory instead
  * remove dependency of sequencing mode for rule for `FastQC`; now runs separately for each strand
* Changes related to MultiQC and output file/directory structure
  * moving and renaming outputs for MultiQC is no longer required
  * code to create MultiQC custom config externalized into script `scripts/rhea_multiqc_config.py`
  * add MultiQC output files with deterministic output to md5 sum checks performed during execution of `tests/test_integration_workflow/test.{local,slurm}.sh`
  * output filenames for each rule now follow this general structure: `samples/{sample_name}/{rule}/{output_file}`
  * change log directory structure matches results directory structure
* Miscellaneous changes
  * consistent, PEP8-compliant formatting in most parts, including Snakemake files, where allowed
  * remove rule `extract_decoys_salmon`; equivalent file `chrName.txt` produced by `star_index` is used instead
  * add rule `start` which copies sample data to the results directory and enforces uniform naming
  * refactoring of ALFA rules and modification of the CI/CD test to ensure compatibility

Dominik Burri authored 5 years ago and Alex Kanitz committed 5 years ago

- generate nucleotide distribution for unique reads only
- new rule to generate PNG image for MultiQC

BIOPZ-Katsantoni Maria authored 5 years ago and Alex Kanitz committed 5 years ago

In labkey_to_snakemake.py fixed the parameters so that there is 3p as well 5p polya
feature for every mate, which can be matched to the -a -g -A and -G options of cutadapt
depending on which is the sense or antisense mate the appropriate variable is populated
and the rest of variables are filled with 'XXXXXXXXXXXX' which leads to no trimming by
cutadapt. The poly-A trimming rules are fixed to contain all -a -g -A -G options.

moved input_files into top-layer test directory for consistency.

corrected removal of test files

corrected md5sum for config.yaml

remove unnecessary file

- renaming bedgraph
- creating ALFA qc plots

removed conda dependence, moved import statement.

included ALFA in finish rule, corrected annotation.gtf and config.yaml, created new .svg

BIOPZ-Iborra de Toledo Paula authored 5 years ago and Alex Kanitz committed 5 years ago

-  Remove files with non-deterministic output from `tests/test_integration_workflow/expected_output.files`
-  Update MD5 sums in `tests/test_integration_workflow/expected_output.md5`
-  Update new workflow DAG and rule graph images

BIOPZ-Katsantoni Maria authored 5 years ago and Alex Kanitz committed 5 years ago

- fixes some functions in `labkey_to_snakemake.py`
- add optional argument for trimming polyA tails; they are trimmed as follows:
  - if mate is sense, oligo-A is added to sample table for `cutadapt` rule to trim
  - if mate is antisense, oligo-T is added to sample table for `cutadapt` rule to trim
  - if option is set to `--trim_polya`, oligo-X stretch is added to sample table and `cutadapt` will not trim

- log and, if workflow is executed on cluster, cluster log directories are explicitly created in `Snakefile`
- location of main log directory can be configured in `config.yaml` (field `log_dir`, previously: `local_log`; requires change in script `labkey_to_snakemake.py` as well as subworkflows as field name is hard-coded there)
- location of cluster log directory can be configured in `cluster.json` (in field `__default__` -> `out`)
- `config.yaml` and `cluster.json` in `tests/input_files` are set such that a directory `logs/` is created in the directory where Snakemake is run (i.e., the directory of each test); cluster logs are stored in a subdirectory `logs/cluster`
- removes instructions to explicitly create log directories from docs and all test scripts
- cleans up main `Snakefile` (apart from Snakemake-specific syntax, now passes `flake8` linter test)

- trap call functionalized through cleanup() function
- function added to all test scripts
- function prints out exit status of last command before trap
- flag `--verbose` added to Snakemake calls in all test scripts
- script tests rename to follow naming convention 'test_script_<script_name>_<script_run_mode>

BIOPZ-Bak Maciej authored 5 years ago and Alex Kanitz committed 5 years ago

- add rule for input preparation (GTF to BED12)
- add rule for TIN score calculation
- update rule graph and DAG image
- update Slurm cluster config

BIOPZ-Katsantoni Maria authored 5 years ago and Alex Kanitz committed 5 years ago

- add script that prepares Snakemake input files 'samples.tsv' and 'config.yaml' from LabKey table
- script either connects to API directly (with '--remote' and related options) or processes a tab-separated LabKey dump file
- add tests for both use cases
- common input files for tests now in 'tests/input_files'
- update all other tests to account for new file locations
- update documentation

BIOPZ-Katsantoni Maria authored 5 years ago and Alex Kanitz committed 5 years ago

- separate organism genome architecture (different input folder)
- change MD5 checksums to match the new output

CJHerrmann authored 5 years ago and Alex Kanitz committed 5 years ago

- add script `tests/test_rule_graph/test.sh` to generate a rule graph in `images/rule_graph.svg`
- display rule graph created in `README.md` instead of specific workflow DAG
- add test script to GitLab CI config
- renamed test to create workflow DAG from `test_create_dag_chart` to `test_create_dag_image` (also output file is renamed from `images/workflow_dag.svg` to `images/dag_test_workflow.svg`

add tests to workflow integration test `tests/test_workflow_integration` that
- verify that STAR alignments match expected alignments (based on "ground truth" files)
- verify that Salmon gene quantification assign the correct number of reads to each gene (based on "ground truth files)

resolves #49

- replaces existing larger libraries and annotations in test cases `test_create_dag_chart` and `test_integration_workflow`
- adds the following new test files:
  - `chr1-10000-20000.fa`: artificial chromosome of length 10'000 (based on human chromosome 1)
  - `chr1-10000-20000.gtf`: matching gene annotation file with two gene and three multi-exon transcripts entries
  - `chr1-10000-20000.transcripts.fa`: sequences of the transcripts listed in the gene annotation file
  - `synthetic.mate_?.fastq.gz`: 10 read pairs randomly sampled from the genic regions of the artificial chromosome
  - `synthetic.*.bed`: BED files with expected alignments for each read; names of overlapping genes are specified in a 7th column
- updates file paths in the relevant sample tables
- extends and updates checksum checking of result files in CI/CD pipeline