vanheeringen-lab / seq2science Goto Github PK

If the bwa mapping gets interrupted, the sort process leaves *tmp.*.bam files. Either sort needs to be updated, or these files need to be deleted before a run starts.

Based on bechmark results?
#5

example: GSM352203

wait for a new release, see FelixKrueger/TrimGalore#64

Fix bug at the same time: #23

related: #8 #7

• samtools index can be run on the bam file that is generate by the rule mark_duplicates

https://github.com/vanheeringen-lab/snakemake-workflows/blob/7946b5b6bd36dd8e07ccd810b29a1d7a6d7d2bca/rules/get_fastq.smk#L2

• In this rule, it might be worth adding a few extra checks and print/echo statements for logging purposes. If this rule fails it is almost always not clear why and at which point it fails.
• Maybe add a retry for esearch or equery if these tools fail due to a timeout?

See for instance GSM1127077, which links to https://www.ncbi.nlm.nih.gov/sra?term=SRX1157701. This is only accessible via dbGap (which we should not bother supporting at the moment).

Get started!

An elegant way of incorporating the installation of ascp:
https://github.com/vanheeringen-lab/snakemake-workflows/blob/54f7e5aaca2f410d120a8a1fcace7a4d18797102/rules/get_fastq.smk#L1-L16

Currently the ascp path and the key are hardcoded.

options:

• install ascp manually, and make sure the hard-coded paths are correct (current situation)
• install ascp manually, do a lookup
  • lookup once per rule
  • lookup once per workflow (e.g. in onstart)
• install ascp through a rule, so the hardcoded paths are enforced

Get it working

We could add for completeness base pair shift
https://www.biostars.org/p/187204/#187206

• star
• hisat2

MACS2 throws away half of the reads when using BAM mode. When using BAMPE the reads get 'interpolated' in between, which we do not want for ATAC-seq.

Ideally align with paired end data, but for peak calling remove the information that the reads are paired end? RNEXT flag in bam

#14 related

Rules using a variable amount of threads should max this out if run on e.g. cartesius.

Ideally it would also defaults when the value is missing

Will scale better with many workflows

Currently only supported for 2 replicates, maybe PR?

Might be confusing to see all the parameters that do not apply. Maybe config.schema.yaml per workflow

When no log specified, it gets auto set:
https://github.com/vanheeringen-lab/snakemake-workflows/blob/a58efbeb0cf5d7b92e669029a280f2c924d83709/schemas/config.schema.yaml#L27-L30

Local files can not be found since they might not be in result_dir, or fastq_dir, or their fqsuffix and/or fqext is wrong. The error

Checking if samples are single-end or paired-end...
CalledProcessError in line 59 of /home/sande/Dropbox/Studie/PhD/snakemake-workflows/rules/configuration.smk:
Command 'esearch -api_key ba36a74749126e0d9558b7e19967417c3407 -db sra -query GSM12345555555 | efetch -api_key ba36a74749126e0d9558b7e19967417c3407 | grep -Po "(?<=<LIBRARY_LAYOUT><)[^/><]*"' returned non-zero exit status 1.
  File "/home/sande/Dropbox/Studie/PhD/snakemake-workflows/workflows/atac_seq/Snakefile", line 11, in <module>
  File "/home/sande/Dropbox/Studie/PhD/snakemake-workflows/rules/configuration.smk", line 78, in <module>
  File "/home/sande/Dropbox/Studie/PhD/snakemake-workflows/rules/configuration.smk", line 78, in <dictcomp>
  File "/home/sande/anaconda3/envs/snakemake-workflows/lib/python3.7/multiprocessing/pool.py", line 657, in get
  File "/home/sande/anaconda3/envs/snakemake-workflows/lib/python3.7/multiprocessing/pool.py", line 121, in worker
  File "/home/sande/Dropbox/Studie/PhD/snakemake-workflows/rules/configuration.smk", line 59, in get_layout
  File "/home/sande/anaconda3/envs/snakemake-workflows/lib/python3.7/subprocess.py", line 395, in check_output
  File "/home/sande/anaconda3/envs/snakemake-workflows/lib/python3.7/subprocess.py", line 487, in run

is completely uninformative

Move https://github.com/vanheeringen-lab/atac-seq to snakemake-workflows!

If the multiqc output files already exist (for instance, from a previous test run), then it will automatically create files with a _1 suffix. However, this means the workflow will fail as the "correct" files according to snakemake are not generated.

Get started!

Get started!

As small as possible reference genomes & samples. This will make testing easier/faster.

Only support for two replicates. Take a look at ENCODE pipeline how they fix it