A Gene Finder framework for Julia.

This is a species-agnostic, algorithm extensible, sequence-anonymous (genome, metagenomes) gene finder library framework for the Julia Language.

The GeneFinder package aims to be a versatile module that enables the application of different gene finding algorithms to the BioSequence type, by providing a common interface and a flexible data structure to store the predicted ORFI or genes. The package is designed to be easily extensible, allowing users to implement their own algorithms and integrate them into the framework.

You can install GeneFinder from the julia REPL. Press ] to enter pkg mode, and enter the following command:

add GeneFinder

The main package function is findorfs. Under the hood, the findorfs function is an interface for different gene finding algorithms that can be plugged using the finder keyword argument. By default it uses the NaiveFinder algorithm, which is a simple algorithm that finds all (non-outbounded) ORFIs in a DNA sequence (see the NaiveFinder documentation for more details).

Here is an example of how to use the findorfs function with the NaiveFinder algorithm:

using BioSequences, GeneFinder

# > 180195.SAMN03785337.LFLS01000089 -> finds only 1 gene in Prodigal (from Pyrodigal tests)
seq = dna"AACCAGGGCAATATCAGTACCGCGGGCAATGCAACCCTGACTGCCGGCGGTAACCTGAACAGCACTGGCAATCTGACTGTGGGCGGTGTTACCAACGGCACTGCTACTACTGGCAACATCGCACTGACCGGTAACAATGCGCTGAGCGGTCCGGTCAATCTGAATGCGTCGAATGGCACGGTGACCTTGAACACGACCGGCAATACCACGCTCGGTAACGTGACGGCACAAGGCAATGTGACGACCAATGTGTCCAACGGCAGTCTGACGGTTACCGGCAATACGACAGGTGCCAACACCAACCTCAGTGCCAGCGGCAACCTGACCGTGGGTAACCAGGGCAATATCAGTACCGCAGGCAATGCAACCCTGACGGCCGGCGACAACCTGACGAGCACTGGCAATCTGACTGTGGGCGGCGTCACCAACGGCACGGCCACCACCGGCAACATCGCGCTGACCGGTAACAATGCACTGGCTGGTCCTGTCAATCTGAACGCGCCGAACGGCACCGTGACCCTGAACACAACCGGCAATACCACGCTGGGTAATGTCACCGCACAAGGCAATGTGACGACTAATGTGTCCAACGGCAGCCTGACAGTCGCTGGCAATACCACAGGTGCCAACACCAACCTGAGTGCCAGCGGCAATCTGACCGTGGGCAACCAGGGCAATATCAGTACCGCGGGCAATGCAACCCTGACTGCCGGCGGTAACCTGAGC"

orfs = findorfs(seq, finder=NaiveFinder) # use finder=NaiveCollector as an alternative

12-element Vector{ORFI{4, NaiveFinder}}:
 ORFI{NaiveFinder}(29:40, '+', 2)
 ORFI{NaiveFinder}(137:145, '+', 2)
 ORFI{NaiveFinder}(164:184, '+', 2)
 ORFI{NaiveFinder}(173:184, '+', 2)
 ORFI{NaiveFinder}(236:241, '+', 2)
 ORFI{NaiveFinder}(248:268, '+', 2)
 ORFI{NaiveFinder}(362:373, '+', 2)
 ORFI{NaiveFinder}(470:496, '+', 2)
 ORFI{NaiveFinder}(551:574, '+', 2)
 ORFI{NaiveFinder}(569:574, '+', 2)
 ORFI{NaiveFinder}(581:601, '+', 2)
 ORFI{NaiveFinder}(695:706, '+', 2)

The ORFI structure displays the location, frame, and strand, but currently does not include the sequence per se. To extract the sequence of an ORFI instance, you can use the sequence method directly on it, or you can also broadcast it over the orfs collection using the dot syntax .:

sequence.(orfs)

12-element Vector{LongSubSeq{DNAAlphabet{4}}}:
 ATGCAACCCTGA
 ATGCGCTGA
 ATGCGTCGAATGGCACGGTGA
 ATGGCACGGTGA
 ATGTGA
 ATGTGTCCAACGGCAGTCTGA
 ATGCAACCCTGA
 ATGCACTGGCTGGTCCTGTCAATCTGA
 ATGTCACCGCACAAGGCAATGTGA
 ATGTGA
 ATGTGTCCAACGGCAGCCTGA
 ATGCAACCCTGA

Similarly, you can extract the amino acid sequences of the ORFIs using the translate function.

translate.(orfs)

12-element Vector{LongAA}:
 MQP*
 MR*
 MRRMAR*
 MAR*
 M*
 MCPTAV*
 MQP*
 MHWLVLSI*
 MSPHKAM*
 M*
 MCPTAA*
 MQP*

ORFIs sequences can be scored using different schemes that evaluate them under a biological context. There are two ways to make this possible: by adding a scoring method to the finder algorithm or by using a scoring method after predicting the ORFIs. The first approach is likely more efficient, but the second approach is more flexible. We will showcase the second approach in this example.

A commonly used scoring scheme for ORFIs is the log-odds ratio score. This score is based on the likelihood of a sequence belonging to a specific stochastic model, such as coding or non-coding. The BioMarkovChains package provides a log_odds_ratio_score method (currently imported), also known as lors, which can be used to score ORFIs using the log-odds ratio approach.

orfs = findorfs(seq, finder=NaiveFinder)

The lors method has been overloaded to take an ORFI object and can be used later to calculate the score of the ORFIs.

lors.(orfs)

12-element Vector{Float64}:
 0.469404606944017
 1.0174520899042823
 1.5914902556997463
 0.9772187907841964
 0.6106494455192994
 0.7089167973379216
 0.469404606944017
 1.5523291911446804
 0.5282685400427601
 0.6106494455192994
 0.7405746713921604
 0.469404606944017

We can extend basically any method that scores a BioSequence to score an ORFI object. To see more about scoring ORFIs, check out the Scoring ORFIs section in the documentation.

GeneFinder also now facilitates the generation of FASTA, BED, and GFF files directly from the found ORFIs. This feature is particularly useful for downstream analysis and visualization of the ORFIs. To accomplish this, the package provides the following functions: write_orfs_fna, write_orfs_faa, write_orfs_bed, and write_orfs_gff.

Functionality:

The package provides four distinct functions for writing files in different formats:

All these function support processing BioSequences instances. To demonstrate the use of the write_* methods with a BioSequence, consider the following example:

using BioSequences, GeneFinder

# > 180195.SAMN03785337.LFLS01000089 -> finds only 1 gene in Prodigal (from Pyrodigal tests)
seq = dna"AACCAGGGCAATATCAGTACCGCGGGCAATGCAACCCTGACTGCCGGCGGTAACCTGAACAGCACTGGCAATCTGACTGTGGGCGGTGTTACCAACGGCACTGCTACTACTGGCAACATCGCACTGACCGGTAACAATGCGCTGAGCGGTCCGGTCAATCTGAATGCGTCGAATGGCACGGTGACCTTGAACACGACCGGCAATACCACGCTCGGTAACGTGACGGCACAAGGCAATGTGACGACCAATGTGTCCAACGGCAGTCTGACGGTTACCGGCAATACGACAGGTGCCAACACCAACCTCAGTGCCAGCGGCAACCTGACCGTGGGTAACCAGGGCAATATCAGTACCGCAGGCAATGCAACCCTGACGGCCGGCGACAACCTGACGAGCACTGGCAATCTGACTGTGGGCGGCGTCACCAACGGCACGGCCACCACCGGCAACATCGCGCTGACCGGTAACAATGCACTGGCTGGTCCTGTCAATCTGAACGCGCCGAACGGCACCGTGACCCTGAACACAACCGGCAATACCACGCTGGGTAATGTCACCGCACAAGGCAATGTGACGACTAATGTGTCCAACGGCAGCCTGACAGTCGCTGGCAATACCACAGGTGCCAACACCAACCTGAGTGCCAGCGGCAATCTGACCGTGGGCAACCAGGGCAATATCAGTACCGCGGGCAATGCAACCCTGACTGCCGGCGGTAACCTGAGC"

Once a BioSequence object has been created, the write_orfs_fna function proves useful for generating a FASTA file containing the nucleotide sequences of the ORFIs. Notably, the write_orfs* methods support either an IOStream or an IOBuffer as an output argument, allowing flexibility in directing the output either to a file or a buffer. In the following example, we demonstrate writing the output directly to a file.

outfile = "LFLS01000089.fna"

open(outfile, "w") do io
    write_orfs_fna(seq, io, finder=NaiveFinder) # use finder=NaiveCollector as an alternative
end

cat LFLS01000089.fna

>seq id=01 start=29 stop=40 strand=+ frame=2 features=[]
ATGCAACCCTGA
>seq id=02 start=137 stop=145 strand=+ frame=2 features=[]
ATGCGCTGA
>seq id=03 start=164 stop=184 strand=+ frame=2 features=[]
ATGCGTCGAATGGCACGGTGA
>seq id=04 start=173 stop=184 strand=+ frame=2 features=[]
ATGGCACGGTGA
>seq id=05 start=236 stop=241 strand=+ frame=2 features=[]
ATGTGA
>seq id=06 start=248 stop=268 strand=+ frame=2 features=[]
ATGTGTCCAACGGCAGTCTGA
>seq id=07 start=362 stop=373 strand=+ frame=2 features=[]
ATGCAACCCTGA
>seq id=08 start=470 stop=496 strand=+ frame=2 features=[]
ATGCACTGGCTGGTCCTGTCAATCTGA
>seq id=09 start=551 stop=574 strand=+ frame=2 features=[]
ATGTCACCGCACAAGGCAATGTGA
>seq id=10 start=569 stop=574 strand=+ frame=2 features=[]
ATGTGA
>seq id=11 start=581 stop=601 strand=+ frame=2 features=[]
ATGTGTCCAACGGCAGCCTGA
>seq id=12 start=695 stop=706 strand=+ frame=2 features=[]
ATGCAACCCTGA

This could also be done to writting a FASTA file with the nucleotide sequences of the ORFIs using the write_orfs_fna function. Similarly for the BED and GFF files using the write_orfs_bed and write_orfs_gff functions respectively.