dcellwanger / PLISH-ProbeDesigner

Probe Designer for PLISH

(c) Daniel C. Ellwanger, 2018-2019.

About

PLISH Probe Designer facilitates the selection of hybridization probes for the proximity ligation in situ hybridization (PLISH) technology recently pulbished by the Harbury and Desai labs at Stanford University (Elife 2018 Jan 10;7. pii: e30510. doi: 10.7554/eLife.30510). PLISH enables rapid and scalable single-cell spatial-profiling of genes of interest using multiplexed hybridization and signal amplification of target RNA species in a single parallel reaction, and the RNAs are then localized within the target tissue with rapid label-image-erase cycles. Therefore, it is a promising technology to inform and validate data analyses from single-cell RNA-Seq experiments.

PLISH Probe Designer facilitates the selection and design of proper hybridization probes (H-probes) for PLISH. For each candidate probe of a given target transcript, PLISH Probe Designer computes a set of features (e.g., melting temperature, probe specificity and fold), which allows the user to select optimal H-probe sequences. Further, for selected probe sequences, PLISH Probe Designer generates the ready-to-order H-probe sequences containing the required connector circle and common bridge sequences for a set of fluorphores (A488, Cy3, Texas Red, Cy5, and PB405).

This tool has been developed and tested using Unix (macOS Sierra).

News

Jan 19 Version 0.4.0

• Corrected LH and RH assignment
• Probe name contains now start instead of center index
• Minor bugfixes

Aug 18 Version 0.3.2

• Transcript versioning numbers are now considered (e.g., 'NM_204873' or 'NM_204873.2')
• Gene name is extracted from GENCODE gff3 files
• Minor bugfixes

Installation

macOS

Simply download this repository and unpack it. To compute some candidate probe features, PLISH Probe Designer makes use of two external software packages: BLAST+ (Camacho et al., BMC Bioinformatics 2009) and RNAstructure (Reuter and Mathews, BMC Bioinformatics 2010). Please, download both software archives from here and unpack them into the tools folder of PLISH Probe Designer.

Database Creation

To create a transcript database, PLISH Probe Designer requires a gff3 annotation and a matching fasta genome sequence file - the same file types that are commonly used to map RNA-Seq reads. Those files can be obtained from common genome databases, such as ENSEMBL, NCBI, and GENCODE. For consistency reasons, we recommend to use those files that were basis for read alignment and quantification in your single-cell RNA-Seq experiment.

The script createDatabase.py (located in the PLISH Probe Designer directory) allows a convenient creation of a database. Within an active Terminal session, the usage of the script can be shown by:

python createDatabase.py --help

### usage: createDatabase.py [-h] -gff FILEPATH -fna FILEPATH -db ID -name NAME
###                          [-comment COMMENT]
###
### optional arguments:
###  -h, --help        show this help message and exit
###  -gff FILEPATH     annotation GFF file
###  -fna FILEPATH     genome sequence FASTA file
###  -db ID            identifier of database (e.g., mmu_refseq); please, avoid
###                    white-spaces and special characters.
###  -name NAME        name of database
###  -comment COMMENT  any comments to add to the info file (e.g., genome
###                    assembly)

Database Deletion

A database can simply be deleted by removing the respective subfolder in database of the PLISH Probe Designer directory.

Probe Selection

After successful installation, PLISH Probe Designer can be simply started from within an active Terminal:

python probeDesigner.py

Feature Calculation

The first step is to identify all candidate probe sequences and calculate the features. The only information that is needed, is the database and the identifier of the target transcript - its sequence is loaded automatically. After providing this input, hit Run. The status of the computation will be shown in the Progress panel. PLISH Probe Designer automatically runs several thermodynamic analyses (free energy of the canidate probe fold, free energy of the homodimer, and free energy of the duplex with the target region) and a BLAST search against a local organism-specific database to assess probe specificity. Please note that these two steps are quite compute-intensive and therefore, depending on the number of candidates may take some time (~1 minute).

Filter and Export

Next, set the desired parameters to filter proper hybridization probes:

• Minimum GC content
• Minimum probe specificity: isoform-specific (only BLAST hits on the target transcript), gene-specific (BLAST hits also on splice variants of the target gene), non-specific (no BLAST hit with an e-value < 10 or BLAST hits also on transcripts of other genes)
• The desired melting temperature (Tm) range
• The minimum free energy of the RNA fold [the higher the better; a value of 0 kcal/mol denotes no secondary structure]
• The minimum free energy of the homodimer [the higher the better; a value of 0 kcal/mol denotes no homodimerization]
• The maximum free energy of the duplex with the target sequence [the lower the stronger the binding to its target]
• If the probe should span an exon junction site of the target transcript

Then hit the Save button. Two files are written into the results folder: a tab-separated csv file containing all computed features for each probe [can be opened with any Text editor or imported to a Spreadsheet Software, such as MS Excel], and a fna FASTA file [can be opened with any Text editor] containing the ready-to-order H-probe sequences for a set of fluorphores (2X = A488, 3X = Cy3, 4X = Texas Red, 5X = Cy5, and 6X = PB405).

Please note that the number of selected probes can be easily lowered or increased: just adjust the filter parameters and hit Save again. It is not required to re-run the whole feature calculation.

The resulting csv file contains the following columns:

• Hprobe: Id: The identifier of the H-probe composed of 'gene name-transcript id-index'
• Hprobe: Target sequence: The target sequence
• Hprobe: %GC: The GC content of the target sequence
• Hprobe: Multipe exons?: Indicates if the target region spans multiple exons
• Hprobe: Exons: The exons that are spanned by the H-probe
• Hprobe: Specificity: The specificity of the H-probe
• Hprobe: Blast Hits (Ident%): The BLAST hits

For the Left and Right arm of the probe:

• Seq: The sequence of the H-probe arm
• Tm: The melting temperature of the H-probe arm (according to Nakano et. al., PNAS 1999)
• Bimol.: Folding free energy change of homodimerization of the H-probe arm
• Unimol.: Folding free energy change of the self-structure of the H-probe arm
• Duplex: Folding free energy change for the duplex formation of the H-probe arm with the target sequence
• Open5: Energy cost of opening the two base pairs at the 5'-end of the H-probe arm in a duplex with the complementary sequence
• Open3: Energy cost of opening the two base pairs at the 3'-end of the H-probe arm in a duplex with the complementary sequence

Example

In this example, we generate H-probes for the gene TECTA as annotated in the chicken genome (Gallus gallus) by NCBI Genome.

Database Creation

First, we need to download and unpack the gff3 and fasta files. In this example, the files are named GCF_000002315.5_GRCg6a_genomic.gff and GCF_000002315.5_GRCg6a_genomic.fna and are, for example, located in the folder /Users/dcellwanger/Downloads/.

Then, the database is created within a Terminal by:

python createDatabase.py \
-gff /Users/dcellwanger/Downloads/GCF_000002315.5_GRCg6a_genomic.gff \
-fna /Users/dcellwanger/Downloads/GCF_000002315.5_GRCg6a_genomic.fna \
-db ncbi_gga -name 'Chicken (NCBI)' \
-comment 'Gallus gallus assembly GRCg6a'

### Writing info file...
### Extracting exon info...
### Processed 500000 lines ...
### Processed 1000000 lines ...
### Processed 1500000 lines ...
### Calculating exon lengths ...
### Writing exon file ...
### Writing sequence file ...
### Generating BLAST+ database ...
### 
### 
### Building a new DB, current time: 07/19/2018 23:54:43
### New DB name:   /Users/dcellwanger/PLISH-ProbeDesigner/database/ncbi_gga/ncbi_gga
### New DB title:  /Users/dcellwanger/PLISH-ProbeDesigner/database/ncbi_gga/ncbi_gga.fna
### Sequence type: Nucleotide
### Keep MBits: T
### Maximum file size: 1000000000B
### Adding sequences from FASTA; added 62160 sequences in 3.70513 seconds.
### Generation of database "ncbi_gga" is finished.

Probe Selection

Let's start the PLISH Probe Designer (python probeDesigner.py), select the database 'Chicken (NCBI)' and the TECTA transcript NM_204873. Hit the Run button.

### Target: NM_204873.2 ("TECTA")
### #Candidates: 548
### Step 1/4: Analyzing splice junction sites...
### Step 2/4: Calculating melting temperature...
### Step 3/4: Calculating thermodynamics...
### Step 4/4: Assessing specificity...
### ------------------[ DONE ]------------------

Then, Save the probes using the standard filter settings. This generates the files TECTA-NM_204873.2_hprobe.csv and TECTA-NM_204873.2_hprobe.fna for 4 selected probes in the results directory. In the latter file, for example, we can then extract the sequences for the H-probe detectable by PB405 (6X):

>HL6X-TECTA-NM_204873.2-5795
TAGGTCAGGAAACTTACGTCGTTATGACGATGTGAGTGCTGTTGGA
>HR6X-TECTA-NM_204873.2-5795
TCCACACCGTGTTCTTGTATTTATACGTCGAGTTGAATAGCCAGGTT

>HL6X-TECTA-NM_204873.2-5938
TAGGTCAGGAAACTTACGTCGTTATGTGAGCATTGGCCGCACGACT
>HR6X-TECTA-NM_204873.2-5938
CACTGTCAGGTTGATCACACTTATACGTCGAGTTGAATAGCCAGGTT

>HL6X-TECTA-NM_204873.2-6169
TAGGTCAGGAAACTTACGTCGTTATGAGCGTAGTTTGTCATTGCTG
>HR6X-TECTA-NM_204873.2-6169
CCCTCCCTCAATGATGAAGTTTATACGTCGAGTTGAATAGCCAGGTT

>HL6X-TECTA-NM_204873.2-6455
TAGGTCAGGAAACTTACGTCGTTATGTCACACCAGTCAGATCGTTT
>HR6X-TECTA-NM_204873.2-6455
GCTCACAGCCACCGTTGTCCTTATACGTCGAGTTGAATAGCCAGGTT

The probes' feature details can be assessed in the tab-separated csv file:

License

GNU GPLv3