A collection of scripts that chain together other scripts by preparing and then processing/transforming the data. A.K.A shepherding scripts, script uniters, utility scripts, parent scripts, shepherds, control scripts, chained scripts, linked scripts, processors, conveyance scripts, scripted processing, mini-workflows, script managers, workflow faciliators, workflow aides, scripted workflows, scripted data handlers, daisy-chained scripts, connected scripts, multiple scripts chained together.*

The repository will probably be mainly Python scripts, but could include some shell/Bash ones too. The inaugral scripts that went into this repository were Python and used Python to control things at the command line level as well. In other words, from within Python I generated and executed commands to run as if on shell command line.

As written, many of these rely on the EasyProcess module that is layered on the subprocess module and is found pre-installed at PythonAnywhere. I suggest using them or edited versions there for the easiest experience.

*In naming this repository, I based it on the responses of stevegt and Alastair Kerr as posted here.

Update: in some ways, recent work in Jupyter notebooks gluing together different scripts is reminiscent of some of what is accomplished in parts of these mini-pipelines, particularly mixing shell and python tasks, and is a useful alternative, see here and here for links to just some nucleic acid analyses and structure/function analyses examples, respectively.

Update: Keep in mind that Python's functools.partial() may be useful when dealing with differing argument number needs at different points in such a pipeline, see here.

Description of each script

• shepherds_all_list_combinations_thru_find_overlap_in_list.py An accessory script for working with the script find_overlap_in_lists.py when you have several lists and want to find the overlap in all the combinations.

• shepherds_many_lists_thru_identify_duplicates_in_a_list.py
  title speaks for now, until I fix this description

• shepherds_all_list_combinations_thru_find_overlap_in_lists_and_sys_name_to_std.py
  title speaks for now, until I fix this description

• shehperds_all_list_combinations_thru_find_overlap_in_lists_with_Venn_and_sys_name_to_std.py
  title speaks for now, until I fix this description

• shepherds_many_lists_thru_identify_duplicates_in_a_list_and_sys_name_to_std.py
  title speaks for now, until I fix this description

• shepherds_chr_thru_plot_expression_across_chromosomes.py
  Uses the script plot_expression_across_chromosomes.py to analyze the entire yeast genome and then all chromosomes for several samples. Producing images of the plots for all those. The version of this script new relies on the columns in the summary data given the script to match samples provided in the USER ADJUSTABLE VALUES for this script. There is an associated script, generate_reports_for_genome_and_all_chromosomes_various_samples.py, here that takes the produced images and makes a report.

• generate_reports_for_genome_and_all_chromosomes_various_samples.py
  Uses the output from the script, shepherds_chr_thru_plot_expression_across_chromosomes.py, to make a summary report as a pdf. I run this script on PythonAnywhere, where they have the ReportLab module installed under Python 2.7.

• shepherds_read_starts_at_start_of_origins_thru_to_plotting.py
  TBD fully; example here does last steps to produce plot. Utilizes the --export_data option to run plot_coverage_and_starts.py in an automated manner and plot the starts in a subregion of the data using the script plot_panel_bar_plots_with_fit.py. An example of what it produces is shown here where the asscoiated script is described. The associated script plot_panel_bar_plots_with_fit.py is found here, as it is general and not specific to read data; it might be placed in the directory with shepherds_read_starts_at_start_of_origins_thru_to_plotting.py .

• extract_data_on_line_using_word_listONhardcodedFILESminiPIPELINE.py
  Adaptation of extract_data_on_line_using_word_list.py from my text-mining repo to use two different word lists to extract two sets of data out of each of several data files as a mini-pipeline. Everything is hard-coded into the script as a quicker route to getting the data without regard to generalizing the script. (Of course, that would be useful if I had time or more need for this later; for now this script serves as a good guide to how a quicker route via a mini-pipeline can be achieved hard-coding file references.) The choice to hard-code the file paths in instead of providing them on the command line as a call also obviated the need to escape the spaces in the file paths since I had copied them from Finder on the Mac and had included those. (Note about adapting further: I had occasion to use this script with DESeq2-generated output, and I found changing the last conditional from if any(word.lower() in line.lower().split() for word in list_of_words): to if any(word.lower() in line.lower().split('"') for word in list_of_words): was necessary because DESeq2 has the gene names flanked by quotes.)

Approaches even more "structured" than my mini-pipelines or MAKEFILES or Bash scripts, some REAL pipeline options I have seen for when I outgrow shell scripts and using subprocess/sh/pybash:

• https://twitter.com/VdaGeraldine/status/1349616600098496513 January 2021

"Impressive review, must-read for anyone getting started with #bioinformatics workflows/pipelines"
"Streamlining data-intensive biology with workflow systems. #Workflows #DataIntensive #Bioinformatics https://academic.oup.com/gigascience/article/10/1/giaa140/6092773 via @GigaScience"

• snakemake - even makes schematic directed acyclic graphs (DAG) of your workflow, see here
  My various snakemake endeavors, making proper pipelines/workflows that take advantage of a full-fledged workflow management system engine, posted on GitHub can be found by clicking here.
  (also see nbpipeline - Snakemake-like pipelines for Jupyter Notebooks, producing interactive pipeline reports)

• Bioinformatics pipelineing option - NExtFlow and Common Workflow Language. See here "This GUI is for connecting command line tools into workflows. It generates CWL workflow descriptions that are human readable, yes." and here and here for help and discussion.

• FlowCraft is an assembler of pipelines written in nextflow for analyses of genomic data. The premisse is simple: Software are container blocks → Build your lego-like pipeline → Execute it (almost) anywhere.

• The Broad Institute's Workflow Description Language( WDL) User Guide

• see here for some comparison in 2019 of several of the above items.

• Ploomber - pointed out here as an option for Notebook-based Workflows. "Ploomber allows you to modularize your analysis [pipelines], making your project more maintainable and easier to test. Since it integrates with Jupyter, you don't have to compromise interactivity, you can still use notebooks and build a production-ready pipeline. Tasks are not limited to notebooks, they can also be scripts or even Python functions."

• Carpentries page on Automation and Make

"Make is used to compile source code into executable programs or libraries, but Make can also be used to:
run analysis scripts on raw data files to get data files that summarize the raw data;
run visualization scripts on data files to produce plots; and to
parse and combine text files and plots to create papers."

• Luigi- Luigi is a Python (2.7, 3.3, 3.4, 3.5, 3.6) package that helps you build complex pipelines of batch jobs. It handles dependency resolution, workflow management, visualization, handling failures, command line integration, and much more. Looks like it is what Spotify uses.

• Pypeline

"Pypeline is a simple yet powerful python library for creating concurrent data pipelines. Pypeline was designed to solve simple medium data tasks that require concurrency and parallelism but where using frameworks like Spark or Dask feel exaggerated or unnatural."

• Tibanna: open-source software for automated execution of bioinformatics pipelines on Amazon Web Services (AWS), article about it here

• snakepipes: facilitating flexible, scalable and integrative epigenomic analysis (July 2019) code https://github.com/maxplanck-ie/snakepipes
  docs https://snakepipes.readthedocs.io/en/latest/

"Due to the rapidly increasing scale and diversity of epigenomic data, modular and scalable analysis workflows are of wide interest. Here we present snakePipes, a workflow package for processing and downstream analysis of data from common epigenomic assays: ChIP-seq, RNA-seq, Bisulfite-seq, ATAC-seq, Hi-C and single-cell RNA-seq. "

• Bioinformatics Pipeline using JUDI: Just Do It (July 2019)

"We developed JUDI on top of a Python based WMS, DoIt, for a systematic handling of pipeline parameter settings based on the principles of DBMS that simplifies plug-and-play scripting. The effectiveness of JUDI is demonstrated in a pipeline for analyzing large scale HT-SELEX data for transcription factor and DNA binding where JUDI reduces scripting by a factor of five.