Design and implement `fastq2count` module

[abearab]

• [x] Discuss problem definition with @nick-youngblut
• [x] Perform exploratory and test data analysis on existing datasets
  • This repo
• [x] Define the final algorithm
• [x] Develop functions and modules
  • [ ] #70
  • [ ] #47
  • [x] https://github.com/wheretrue/biobear/issues/89
  • I had to install biobear with pip instead of conda to make sure I have most recent version. https://github.com/ArcInstitute/ScreenPro2/commit/9180855221b53fff993a02dac74e3ba05c9ee01e
  • [x] https://github.com/wheretrue/biobear/issues/96
  • [x] https://github.com/wheretrue/biobear/issues/102

    ---

    related tools / scripts:

• @noamteyssier tools - sgcount, casmap
• @mhorlbeck scripts – fastqgz_to_counts.py
• scripts for processing dual-sgRNA-screens – CRISPRi-dual-sgRNA-screens/module1
• "ReCo: automated NGS read-counting of single and combinatorial CRISPR gRNAs" | Paper
• "pgMAP: a pipeline to enable guide RNA read mapping from dual-targeting CRISPR screens" | arXiv
• guide-counter

[abearab]

PosSpacer – https://github.com/abearab/PosSpacer [archived]

---

old notes ...

A flexible NGS read counter for CRISPR screens

Inputs: (1) FASTQ files, (2) Sample sheet (3) Oligo pool reference sequences Outputs: (1) Oligo count matrix, (2) Count level QC reports

• Preprocess FASTQ files → Count occurrences

• Assemble reads based on the overlap of R1 and R2 [optional]

• Define rules to clean up reads based on the library prep protocol and oligo library design strategy (e.g. trimming task, etc.)

• Find all unique sequences and count occurrences

• Map to reference library → Count matrix

• Map observed sequences to perfect match in the reference library

Expected features:

• Compatibility with multiple CRISPR screen platforms
• Optimized codes for pipeline development
• Documentation for simple usage by end-users
• Count level QC reports (e.g. replicate correlations, count distribution histograms)
• Export side tables for unexpected oligo fragments [novel]