prbdes 0.1.2

Probe design originails for FISH by Quentin Verron

Instructions for probe design

Installation

• Install the dev branch of nHUSH

• Install HUSH

• Install escafish

• Install OligoArrayAux

• Install ifpd2

• Install oligo-melting

Notes on ifpd2 installation:

If ifpd2 is already installed, remove using pip3 uninstall ifpd2.

Get a new copy of the ifpd2 repository: git clone https://github.com/ggirelli/ifpd2.git

Manually edit ifpd2/const.py: Row 16: dtype_hush={"sequence":"|S","off_target_no":"<u8"}

Manually edit ifpd2/io.py: Row 66: ass.ert_in_dtype(hush_df["off_target_no"].values.max(),"u8")

Install using (requires poetry):

cd ifpd2
poetry install
poetry build
pip3 install dist/*.whl

Preparation

DNA:

• Get the genomic coordinates of the regions of interest

• Get the reference genome

RNA:

• Get the transcripts of interest

• Get the reference transcriptome

Notes:

• For DNA probes, the reference genome will be used both to extract the sequences of interest and to test probe candidates for homology. If different genomes need to be used, follow RNA steps and provide the regions of interest directly.

• For combined DNA-RNA FISH, the probe sets should be designed with an homology check against both genome and transcriptome.

• All the commands below assume you are starting from your pipeline installation folder:

  cd probe_design
  

Probe design pipeline:

Alternative 1: Normally repetitive regions.

1. Preparation

• The probe desin pipeline data is currently intended to be run on a folder called data/ contained within the pipeline installation folder.

• Upon starting the pipeline, the data/ folder should only contain data/rois/ and data/ref/ (and possibly data/blacklist/, see 6.). If more folders are included, consider making a back-up or simply removing them.

• List your regions of interest and their coordinates in the input file: data/rois/all_regions.tsv

• Place your reference genome in the data/ref/ folder. Make sure that the chromosome naming matches with the reference genome name provided in all_regions.tsv.

• The reference folder can alternatively be gathered using get_ref_genome.sh. In that case, adjust the script manually with the correct Ensembl address for your genome of interest.

2. Generate all required subfolders:

mkdir data/candidates
mkdir data/melt
mkdir data/secs
mkdir data/db
mkdir data/db_tsv
mkdir data/logfiles
mkdir HUSH

3. Retrieve your region sequences and extract all k-mers of correct length:

   # (from Pipeline/)
   ./get_oligos.py DNA|RNA [optional: applyGCfilter 0|1]
   # Example:
   ./get_oligos.py DNA 1
   

   If indicating RNA, the module will assume that the transcript / region sequences are already present in the data/regions folder. Default: `DNA.

4. Test all k-mers for their homology to other regions in the genome, using nHUSH. Instead of running the entire k-mers (of length L) at once, can be sped up by testing shorter sublength oligos (of length l). -m number of mismatches to test for (always use 1 when running sublength); -t number of threads, -i comb size

• Full length:

  ./run_nHUSH.sh -d RNA -L 35 -m 5 -t 40 -i 14
  

• Sublength:

  ./run_nHUSH.sh -d DNA -L 40 -l 21 -m 3 -t 40 -i 14
  

  ADD -g if this is the first time running with a new reference genome!

• In case nHUSH is interrupted before completion, run before continuing:

  ./unfinished_HUSH.sh
  

  5. Recapitulate nHUSH results as a score

./reform_hush_combined.py DNA|RNA|-RNA length sublength until

(until denotes the same number as specified after -m when running nHUSH).

6. Calculate the melting temperature of k-mers and the free energy of secondary structure formation:

   ./melt_secs_parallel.sh (optional DNA(ref) / RNA(rev. compl))   
   

7. Generate a black list of abundantly repeated oligos in the reference genome.

   ./generate_blacklist.sh -L 40 -c 100
   

   This only needs to be run once per reference genome if not using any exclusion regions! Just save the blacklist folder between runs.

L: oligo length; c: min abundance to be included in oligo black list

8. Create k-mer database, convert to TSV for querying and attribute score to each oligo (based on nHUSH score, GC content, melting temperature, homopolymer stretches, secondary structures).

    ./build-db_BL.sh -f q_bl -m 32 -i 6 -L 40 -c 100 -d 8 -T 72
   

   m: Maximum length of a consecutive match. Default: 24 i: Maximum length of a consecutive homopolymer. Default: 6 All oligos with a longer consecutive match or homopolymer are stricly excluded. L: oligo length c: min number of occurrences for an oligo to be counted in black list (should match settings used in 6.) d: min Hamming distance to an oligo in the blacklist for exclusion T: Target melting temperature. Default: 72C

9. Query the database to get candidate probes:

   ./cycling_query.py -s DNA -L 40 -m 8 -c 100 -t 40 -greedy
   

    [optional: -greedy. Speed > quality]
   

   [optional: -start 20 -end 100 -step 5] To sweep different oligo numbers, otherwise uses the oligo counts provided in ./rois/all_regions.tsv [optional: -stepdown 10] Number of oligos to decrease probe size with every iteration that does not find enough oligos. Default: 1

Cycling query which generate probe candidates, then checks the resulting oligos using HUSH, removes inacceptable oligos and generate probes again. If enough oligos cannot be found, design probes with fewer oligos, decreasing with stepdown at each step.

10. Summarize the final probes:

./summarize-probes-final.py

Some visual elements can be obtained using the following notebooks (needs updating!):

``` shell
plot_probe_candidates.ipynb
plot_oligos.ipynb
```

Alternative 2: Repetitive or repeated regions.

In this alternative, the region (along with any user-indicated repeats) is masked out from the reference genome used by nHUSH. This way, repeated oligos that are specific for the ROI can be included in the final probe.

Warning: This approach occupies a lot more hard drive space!

1. Preparation

• Besides data/rois/ and data/ref/, the pipeline requires an additional data/exclude/ folder containing BED files with the coordinates of sections to mask out when running HUSH for each ROI.

2. (UNLESS manually providing exclusion regions) Exclude regions of interest from HUSH scan.

./generate_exclude.py

• The same sheet template can be used to manually add further regions to exclude.

2. Generate all required subfolders:

mkdir data/candidates
mkdir data/melt
mkdir data/secs
mkdir data/db
mkdir data/db_tsv
mkdir data/logfiles

3. Retrieve your region sequences and extract all k-mers of correct length:

   # (from Pipeline/)
   ./get_oligos.py DNA|RNA [optional: applyGCfilter 0|1]
   # Example:
   ./get_oligos.py DNA
   

   If indicating RNA, the module will assume that the transcript / region sequences are already present in the data/regions folder. Default: `DNA.

4. Apply the region exclusion mask on the reference genome.

   ./exclude_region.py 
   

5. Generate a black list of abundantly repeated oligos in the reference genome.

   ./generate_blacklist.sh -L 40 -c 100
   

Needs to be re-run everytime when using exclusion masks. L: oligo length; c: min abundance to be included in oligo black list

6. Test all k-mers for their homology to other regions in the genome, using nHUSH. Instead of running the entire k-mers (of length L) at once, can be sped up by testing shorter sublength oligos (of length l). -m number of mismatches to test for (minimum 1 for sublength; more gives better information but takes longer time); -t number of threads, -i comb size

Sublength:

./run_nHUSH_excl.sh -d DNA -L 40 -l 21 -m 3 -t 40 -i 14

Note the _excl specific to the exclusion mode.

In case nHUSH is interrupted before completion, run before continuing:

./unfinished_HUSH.sh

7. Recapitulate nHUSH results as a score

# Format:
./reform_hush_combined.py DNA|RNA|-RNA length sublength until
# Example:
./reform_hush_combined.py DNA 40 21 3

(until denotes the same number as specified after -m when running nHUSH).

8. Calculate the melting temperature of k-mers and the free energy of secondary structure formation:

   ./melt_secs_parallel.sh (optional DNA(ref) / RNA(rev. compl))
   

9. Create k-mer database, convert to TSV for querying and attribute score to each oligo (based on nHUSH score, GC content, melting temperature, homopolymer stretches, secondary structures).

   Recommended:

    ./build-db_BL.sh -f q_bl -m 32 -i 6 -L 40 -c 100 -d 8 -T 72
   

   f: score function d: max Hamming distance to blacklist that is excluded L: oligo length; c: min abundance to be included in oligo blacklist i: max identical consecutive base pairs, T: target temperature, m: max length of consecutive off-target match

10. Query the database to get candidate probes:

    ./cycling_query.py -s DNA -L 40 -m 8 -c 100 -t 40 -g 500 -stepdown 50 -greedy -excl
    

    [optional: -greedy. Speed > quality]
    

    [optional: -start 20 -end 100 -step 5] To sweep different oligo numbers, otherwise uses the oligo counts provided in ./rois/all_regions.tsv [optional: -stepdown 10] Number of oligos to decrease probe size with every iteration that does not find enough oligos. Default: 1

Cycling query which generate probe candidates, then checks the resulting oligos using HUSH, removes inacceptable oligos and generate probes again. If enough oligos cannot be found, design probes with fewer oligos, decreasing with stepdown at each step.

11. Summarize the final probes:

python summarize-probes-final.py

Generate probes for ordering

• Select forward, reverse primers and color flaps.
• Add the forward and reverse primer sequences to the probe oligos
• The forward primer to order has the color flap + the forward sequence
• The reverse primer to order has the t7 promoter sequence + the rev. compl of the rev sequence in the oligo
• The complete oligos can be uploaded as an Excel file containing the oligo names (arbitrary but unique) and the sequences

TO DO:

• Adapt the code for more flexibility in input/output folders.
• Add a visual report of the probes at the end of the pipeline.
• One-button process!
• Find a way to automatize selecting primer sequences.