genet 0.7.0

GenET: Genome Editing Toolkit

Genome Editing Toolkit
Since 2022. 08. 19.

Welcome to GenET

GenET (Genome Editing Toolkit) is a library of various python functions for the purpose of analyzing and evaluating data from genome editing experiments. GenET is still in its early stages of development and continue to improve and expand. Currently planned functions include guideRNA design, saturation library design, deep sequenced data analysis, and guide RNA activity prediction.

System requirement

GenET can be run on either Mac or Linux system. GenET is currently available on Linux or Mac based systems as one of the dependent tools, ViennaRNA package, is limited to these operating systems. Windows users must establish a WSL, docker or virtual OS environment to use this tool.

Installation

1/ Create virtual environment and install genet

# Create virtual env for genet. (python 3.8 was tested)
conda create -n genet python=3.8
conda activate genet

# Install genet ( >= ver. 0.7.0)
pip install genet

2/ Install Pytorch (v1.11.0 was tested)

Pytorch ver.2 is not compatible yet.

# For OSX (MacOS)
pip install torch==1.11.0

# For Linux and Windows
# CUDA 11.3
pip install torch==1.11.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113

# CUDA 10.2
pip install torch==1.11.0+cu102 --extra-index-url https://download.pytorch.org/whl/cu102

# CPU only
pip install torch==1.11.0+cpu --extra-index-url https://download.pytorch.org/whl/cpu

3/ Install ViennaRNA

# install ViennaRNA package for prediction module
conda install viennarna

Trouble shooting for installation

1/ GLIBCXX ImportError

ImportError: /lib64/libstdc++.so.6: version `GLIBCXX_3.4.20' not found (required by /home/hkim/.miniconda3/envs/genet/lib/python3.8/site-packages/RNA/_RNA.cpython-38-x86_64-linux-gnu.so)'

If the above error message appears in the process of loading the Vienna RNA, install higher version of 'libgcc' using mamba (see alse).

conda activate genet
conda install -c conda-forge mamba
mamba install libgcc

Who should use GenET?

GenET was developed for anyone interested in the field of genome editing. Especially, Genet can provide aid to those with the following objectives.:

• Develop a quick and easy to design an genome editing experiment for a specific gene.
• Perform genome editing analysis based on sequening data
• Predict the activtiy of specific guideRNAs or all guideRNAs designed for editing a specific product.

Tutorial 1: Predict SpCas9 activity (by DeepSpCas9)

DeepSpCas9 is a prediction model developed to evaluate to indel frequency introduced by sgRNAs at specific target sites mediated by the SpCas9 (Kim et al. SciAdv 2019). The model was developed on tensorflow (version >= 2.6). Any dependent packages will be installed along with the GenET package.

from genet import predict as prd

# Put the target context (30bp) that you want to find Cas9 activity in the list.
# Input seq: 4bp 5' context + 20 guide + 3bp PAM + 3bp 3' context

list_target30 = [
                'TCACCTTCGTTTTTTTCCTTCTGCAGGAGG',
                'CCTTCGTTTTTTTCCTTCTGCAGGAGGACA',
                'CTTTCAAGAACTCTTCCACCTCCATGGTGT',
                ]
                
list_out = prd.spcas9_score(list_target30)

list_out
>>> [2.80322408676147, 2.25273704528808, 53.4233360290527]

Tutorial 2: Predict Prime editing efficiency (by DeepPrime and DeepPrime-FT)

DeepPrime is a prediction model for evaluating prime editing guideRNAs (pegRNAs) that target specific target sites for prime editing (Yu et al. Cell 2023). DeepSpCas9 prediction score is calculated simultaneously and requires tensorflow (version >=2.6). DeepPrime was developed on pytorch.

from genet import predict as prd

# Place WT sequence and Edited sequence information, respectively.
# And select the edit type you want to make and put it in.
#Input seq: 60bp 5' context + 1bp center + 60bp 3' context (total 121bp)

seq_wt   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
seq_ed   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
alt_type = 'sub1'

df_pe = prd.pe_score(seq_wt, seq_ed, alt_type)
df_pe.head()

output:

	ID	WT74_On	Edited74_On	PBSlen	RTlen	RT-PBSlen	Edit_pos	Edit_len	RHA_len	type_sub	type_ins	type_del	Tm1	Tm2	Tm2new	Tm3	Tm4	TmD	nGCcnt1	nGCcnt2	nGCcnt3	fGCcont1	fGCcont2	fGCcont3	MFE3	MFE4	DeepSpCas9_score	PE2max_score
0	Sample	ATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATG	xxxxxxxxxxxxxxCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGxxxxxxxxxxxxxxxxxx	7	35	42	34	1	1	1	0	0	16.191	62.1654	62.1654	-277.939	58.2253	-340.105	5	16	21	71.4286	45.7143	50	-10.4	-0.6	45.9675	0.0202249
1	Sample	ATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATG	xxxxxxxxxxxxxCCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGxxxxxxxxxxxxxxxxxx	8	35	43	34	1	1	1	0	0	30.1995	62.1654	62.1654	-277.939	58.2253	-340.105	6	16	22	75	45.7143	51.1628	-10.4	-0.6	45.9675	0.0541608
2	Sample	ATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATG	xxxxxxxxxxxxACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGxxxxxxxxxxxxxxxxxx	9	35	44	34	1	1	1	0	0	33.7839	62.1654	62.1654	-277.939	58.2253	-340.105	6	16	22	66.6667	45.7143	50	-10.4	-0.6	45.9675	0.051455
3	Sample	ATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATG	xxxxxxxxxxxCACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGxxxxxxxxxxxxxxxxxx	10	35	45	34	1	1	1	0	0	38.5141	62.1654	62.1654	-277.939	58.2253	-340.105	7	16	23	70	45.7143	51.1111	-10.4	-0.6	45.9675	0.0826205
4	Sample	ATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATG	xxxxxxxxxxACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGxxxxxxxxxxxxxxxxxx	11	35	46	34	1	1	1	0	0	40.8741	62.1654	62.1654	-277.939	58.2253	-340.105	7	16	23	63.6364	45.7143	50	-10.4	-0.6	45.9675	0.0910506

It is also possible to predict other cell lines (A549, DLD1...) and PE systems (PE2max, PE4max...).

from genet import predict as prd

seq_wt   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
seq_ed   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
alt_type = 'sub1'

df_pe = prd.pe_score(seq_wt, seq_ed, alt_type, sID='MyGene', pe_system='PE4max', cell_type='A549')

Tutorial 3: Get ClinVar record and DeepPrime score using GenET

ClinVar database contains mutations that are clinically evaluated to be pathogenic and related to human diseases(Laudrum et al. NAR 2018). GenET utilized the NCBI efect module to access ClinVar records to retrieve related variant data such as the genomic sequence, position, and mutation pattern. Using this data, genET designs and evaluates pegRNAs that target the variant using DeepPrime.

from genet import database as db

# Accession (VCV) or variantion ID is available
cv_record = db.GetClinVar('VCV000428864.3')

print(cv_record.seq()) # default context length = 60nt

>>> output: # WT sequence, Alt sequence
('GGTCACTCACCTGGAGTGAGCCCTGCTCCCCCCTGGCTCCTTCCCAGCCTGGGCATCCTTGAGTTCCAAGGCCTCATTCAGCTCTCGGAACATCTCGAAGCGCTCACGCCCACGGATCTGC',
 'GGTCACTCACCTGGAGTGAGCCCTGCTCCCCCCTGGCTCCTTCCCAGCCTGGGCATCCTTGTTCCAAGGCCTCATTCAGCTCTCGGAACATCTCGAAGCGCTCACGCCCACGGATCTGCAG')

In addition, various information other than the sequence can be obtained from the record.

# for example, variant length of the record
print(cv_record.alt_len)

>>> output:
2

Clinvar records obtained through this process is used to design all possible pegRNAs within the genet.predict module's pecv_score function.

from genet import database as db
from genet import predict as prd

cv_record = db.GetClinVar('VCV000428864.3')
prd.pecv_score(cv_record)

Tutorial 4: Make additional synonymous mutations in pegRNA (GenET design module)

from genet import predict
from genet import design

seq_wt   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGAAGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
seq_ed   = 'ATGACAATAAAAGACAACACCCTTGCCTTGTGGAGTTTTCAAAGCTCCCAGAAACTGAGACGAACTATAACCTGCAAATGTCAACTGAAACCTTAAAGTGAGTATTTAATTGAGCTGAAGT'
alt_type = 'sub1'

df_pe = predict.pe_score(seq_wt, seq_ed, alt_type)

# Select pegRNA that you want to add synonymous mutation 
# The record type should be pd.Series
dp_record = df_pe.iloc[20]

synony_pegrna = design.SynonymousPE(dp_record, ref_seq=seq_wt, frame=1)

pegrna_ext = synony_pegrna.extension

Tutorial 5: Get Gene information from NCBI (GenET database module)

The database module is used to retrieve sequence and feature information regarding the target gene of interest. This process is based on the Entrez module on biopython. Currently, obtaining only the meta data cooresponding to each feature is available, but in the future, we plan to implement sequence retreival followed by full preprocessing of neccesary information required for genome editing.

ex) Retrieve gene info from NCBI

from genet import database as db
# If you import for the first time, you have to enter an email.
# This is because it is required to leave a log when accessing NCBI's Entrez database.

brca1 = db.GetGene('BRCA1')

list_exons = brca1.exons()
list_exons

>>> output:
[SeqFeature(FeatureLocation(ExactPosition(92500), ExactPosition(92713), strand=1), type='exon'),
 SeqFeature(FeatureLocation(ExactPosition(93868), ExactPosition(93967), strand=1), type='exon'),
 SeqFeature(FeatureLocation(ExactPosition(102204), ExactPosition(102258), strand=1), type='exon'),
 SeqFeature(FeatureLocation(ExactPosition(111450), ExactPosition(111528), strand=1), type='exon'),
 SeqFeature(FeatureLocation(ExactPosition(113027), ExactPosition(113116), strand=1), type='exon'),
......
]

Please send all comments and questions to [e-mail]