What is Gene Prediction?

• In computational biology, gene prediction or gene finding refers to the identification of the genomic DNA regions that encode genes.
• This includes both protein-coding and RNA genes, but may also include the prediction of regulatory regions and other functional elements. Once a species’ genome has been sequenced, gene discovery is one of the first and most crucial stages in comprehending its genome.
• Initially, “gene finding” was founded on painstaking experiments conducted on living cells and organisms.
• Statistical analysis of the rates of homologous recombination of several different genes could determine their order on a particular chromosome, and information from many such experiments could be combined to produce a genetic map indicating the approximate relative location of known genes. With a complete genome sequence and potent computational resources at the disposal of the scientific community, gene discovery is now primarily a computational problem.
• Differentiate between determining the function of a gene or its progeny and determining the functionality of a sequence. Although the frontiers of bioinformatics research are making it possible to predict the function of a gene based on its sequence alone, predicting the function of a gene and confirming that the gene prediction is accurate still requires in vivo experimentation through gene knockout and other assays.
• Gene prediction is one of the most important stages in genome annotation, following sequence assembly, non-coding region filtering, and repeat masking.
• The so-called ‘target search problem’ investigates how DNA-binding proteins (transcription factors) locate specific binding sites within the genome. Gene prediction is closely related to this problem.
• Numerous aspects of structural gene prediction are based on the current understanding of biochemical processes in the cell, such as gene transcription, translation, protein–protein interactions, and regulation processes, which are the subject of active research in the various omics fields, including transcriptomics, proteomics, metabolomics, and more generally structural and functional genomics.

Example

Gene: A sequence of nucleotides coding for protein

Gene Prediction Problem: Determine the beginning and end positions of genes in a genome

atgcatgcggctatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatcctgcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagct

Gene

atgcatgcggctatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatcctgcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagctcggctatgctaatgaatggtcttgggatttaccttggaatgctaagctgggatccgatgacaatgcatgcggctatgctaatgaatggtcttgggatttaccttggaatatgctaatgcatgcggctatgctaagctgggaatgcatgcggctatgctaagctgggatccgatgacaatgcatgcggctatgctaatgcatgcggctatgcaagctgggatccgatgactatgctaagctgcggctatgctaatgcatgcggctatgctaagct

Bioinformatics and the Prediction of Genes

Utilising computational algorithms and tools to analyse genomic data, bioinformatics plays an essential role in the prediction of genes. Here are the contributions of bioinformatics to gene prediction:

• Sequence Analysis: Bioinformatics algorithms analyse DNA and RNA sequences to identify potential gene regions during sequence analysis. Identifying essential sequence features, such as promoter regions, transcription start sites, splice sites, and termination signals, is required. BLAST (Basic Local Alignment Search Tool) and other sequence alignment techniques are frequently used to compare and match sequences against known gene sequences or databases.
• Gene Finding Algorithms: Algorithms Bioinformatics tools use various gene location prediction algorithms to predict gene locations within genomic sequences. These algorithms employ statistical models, hidden Markov models (HMMs), or machine learning techniques to identify regions with unique gene signatures, such as coding regions (exons) and non-coding regions (introns). GENSCAN, AUGUSTUS, and GeneMark are well-known gene-finding algorithms.
• Comparative Genomics: Bioinformatics permits comparative analysis of genomic sequences from various species. Researchers can identify conserved regions that correspond presumably to genes by aligning and comparing genomes. Comparative genomics facilitates gene prediction by utilising the principle that functional genes are frequently conserved among closely related species.
• Transcriptomics and RNA-Seq: Bioinformatics tools are widely used to analyse transcriptomic data, such as RNA-Seq, which provides gene expression level information. Bioinformatics algorithms can identify and quantify gene expression levels by mapping RNA-Seq reads to a reference genome, thereby assisting gene prediction. Moreover, transcriptomic data aids in the refinement of gene models, the identification of alternative splicing events, and the detection of noncoding RNA molecules.
• Functional Annotation: Bioinformatics resources and databases facilitate the functional annotation of predicted genes. Using databases such as UniProt, Gene Ontology (GO), and Kyoto Encyclopaedia of Genes and Genomes (KEGG), it is possible to designate functional annotations, predict protein domains, and infer biological functions for predicted gene sequences. Understanding the functions of predicted genes in biological processes is impossible without functional annotation.
• Validation and Experimental Design: Predictions made by bioinformatics are frequently validated by experimental techniques. RT-PCR, for example, can be used to validate predicted gene expression patterns and corroborate the presence of predicted alternative splicing events. Bioinformatics tools also facilitate the design of targeted experiments by providing insights into gene function and regulatory elements.
• Construction and Maintenance of Databases: Bioinformatics contributes to the construction and maintenance of gene and genome databases. These databases contain gene sequence predictions, annotations, and related data. The databases of the National Centre for Biotechnology Information (NCBI), Ensembl, and the UCSC Genome Browser are examples. These databases serve as valuable resources for gene and genome researchers.

Using computational algorithms, comparative genomics, transcriptomic analysis, functional annotation, and database construction, bioinformatics plays a significant role in gene prediction. Incorporating bioinformatics into genomics research enables scientists to predict and comprehend the complexity of gene structures, regulatory elements, and their functional significance.

Methods of Gene Prediction

The process of identifying the locations and boundaries of genes within a genome is known as gene prediction. Understanding the genetic information contained in an organism’s DNA is a crucial step. Among the many methods and algorithms used for gene prediction are the following:

1. Ab initio prediction: These methods predict genes based on the properties of DNA sequences using computational algorithms. They analyse coding potential, sequence motifs, splice sites, and start/stop codons, among other characteristics. The ab initio gene prediction algorithms GeneMark, Fgenesh, and Glimmer are examples.
2. Homology-based prediction: These techniques rely on comparing the DNA sequence to known sequences from organisms with similar characteristics. If a sequence has a high degree of similarity to a known gene, it is likely to be a gene. Homology-based methods seek for similarities using tools such as BLAST (Basic Local Alignment seek Tool). Evolutionarily conserved genes are assumed to have comparable sequences in closely related species.
3. EST-based prediction: ESTs (Expressed Sequence Tags) are brief sequences derived from cDNA libraries that represent segments of expressed genes. EST-based gene prediction entails aligning ESTs to genomic sequences and identifying overlapping regions. This method is especially beneficial when working with organisms for which genomic data is scarce.
4. Transcriptome-based prediction: These methods identify gene regions based on RNA sequencing (RNA-seq) data. RNA-sequencing provides information regarding the transcripts present in a particular tissue or condition. By aligning RNA-seq reads to the genome, gene locations and alternative splicing patterns can be inferred.
5. Comparative genomics: This method entails comparing the genomes of various species to identify conserved regions that are likely to correspond to genes. By aligning genomes, researchers can identify regions that may represent functional genes that are shared between species.
6. Machine learning and deep learning: Using advanced computational techniques, machine learning and deep learning train models on large datasets of known genomes. By recognising patterns within these datasets, the models are able to predict new genomic sequences. Machine learning and deep learning techniques, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), have shown promise in gene prediction tasks.

Ab initio prediction

• Ab initio gene prediction is a computational method for identifying genes based on DNA sequence properties.
• It does not require prior knowledge of gene sequences or similarities to previously identified genes.
• Various algorithms and software applications, such as GeneMark, Fgenesh, and Glimmer, are utilised for ab initio gene prediction.
• Ab initio methods analyse DNA sequence features, such as coding potential, sequence motifs, splice sites, and start/stop codons, to identify potential genes.
• To distinguish coding regions from non-coding regions, these methods frequently employ statistical models and machine learning techniques.
• Ab initio prediction is applicable to both bacterial and eukaryotic genomes.
• Ab initio methods are particularly useful when studying non-model organisms or organisms with limited genomic data.
• Depending on the quality of the underlying models and the complexity of the genome being analysed, the precision of ab initio gene prediction can vary.
• False positives (mistakenly identifying a region as a gene) and false negatives (missing actual genes) are typical difficulties in ab initio gene prediction.
• To enhance accuracy and gene annotations, Ab initio predictions are frequently combined with other methods, such as homology-based or transcriptome-based predictions.
• Manual curation and experimental validation are essential stages for refining ab initio predictions and confirming the existence and function of predicted genes.

Ab initio gene prediction is a useful method for identifying genes in genomic sequences based on intrinsic sequence properties. It serves as a starting point for further research and can aid in the comprehension of the genetic information and functional components of a genome.

Homology-based prediction

• Homology-based gene prediction relies on comparing the DNA sequence of interest to known gene sequences from related organisms.
• The underlying principle is that genes that perform similar functions in different species tend to have conserved sequences.
• The process begins with performing sequence similarity searches using tools such as BLAST (Basic Local Alignment Search Tool) or FASTA against databases containing known gene sequences.
• If a significant similarity is found between the query sequence and a known gene, it suggests that the query sequence may also be a gene.
• Homology-based prediction is particularly useful when studying well-characterized model organisms or when working with closely related species.
• It can also help in identifying genes that have conserved functions across different species.
• The quality of homology-based predictions depends on the completeness and accuracy of the reference gene databases used.
• Gene predictions can be refined by considering additional information such as conservation of gene order (synteny) between species.
• Multiple sequence alignment techniques can be applied to identify conserved regions within homologous genes.
• Homology-based predictions can provide valuable insights into the putative function and structure of the predicted genes.
• However, it may not be effective for identifying genes that are specific to a particular species or those that have undergone significant evolutionary divergence.
• Validation of homology-based predictions often requires experimental evidence such as gene expression studies or functional assays.

EST-based prediction

• Expressed Sequence Tags (ESTs) are brief sequences derived from cDNA libraries that represent portions of expressed genes.
• By sequencing the mRNA molecules present in a specific tissue or condition, ESTs can provide valuable information about the regions of the genome that are transcribed.
• When investigating organisms with limited genomic data or when working with non-model organisms, EST-based prediction can be especially beneficial.
• The procedure begins with the alignment of EST sequences to the genomic DNA sequence using BLAST or other sequence alignment algorithms.
• Potential gene regions are regions of the genome that exhibit significant sequence similarity to ESTs.
• Predictions based on ESTs can aid in the identification of coding regions, untranslated regions (UTRs), alternative splicing events, and other aspects of gene structure.
• By aligning multiple ESTs to the genomic sequence, consensus regions corresponding to exons and introns can be identified.
• The availability and quality of EST databases, as well as the completeness and accuracy of the genomic sequence, influence the quality of EST-based predictions.
• Integration of EST-based predictions with other gene prediction methods can enhance the precision and comprehensiveness of gene annotations.
• Often, experimental validation, such as reverse transcription polymerase chain reaction (RT-PCR) or comparison with known gene structures, is required to affirm the predicted gene boundaries and validate their expression.
• Predictions based on ESTs can provide insightful information about tissue-specific expression patterns and gene regulation.

In conclusion, EST-based gene prediction utilises the information in EST databases to identify potential gene regions and characterise gene structure. It is a beneficial method, especially in the absence of complete genomic data, and it can provide valuable insights into gene expression and regulation. Nevertheless, it is crucial to consider the limitations and potential errors associated with EST-based predictions and to validate the results using experimental methodologies.

Transcriptome-based prediction

• Utilising RNA sequencing (RNA-seq) data, transcriptome-based gene prediction identifies gene regions within a genome.
• RNA-seq provides information about the transcripts present in a particular tissue or condition, enabling the identification of genes that are expressed.
• RNA-seq reads are initially aligned to the genomic DNA sequence using alignment algorithms such as Bowtie, HISAT, or STAR.
• By aligning the RNA-seq sequences, it is possible to identify genomic regions that are transcribed and may correspond to genes.
• In addition to protein-coding genes, alternative splicing events and non-coding RNA genes can be identified through transcriptome-based prediction.
• Gene expression levels can be estimated based on the abundance of RNA-seq reads at specific genomic locations.
• Predictions based on the transcriptome can also aid in the identification of novel genes, isoforms, and fusion genes.
• Multiple replicates and distinct RNA-seq library preparation protocols can be utilised to enhance precision.
• Integration of transcriptome-based gene predictions with other gene prediction approaches, such as ab initio or homology-based methods, can improve gene annotation.
• Experimental validation using methods such as RT-PCR, Northern blotting, or comparison with known gene structures is required to confirm the predicted gene boundaries and validate their expression.
• Gene prediction based on the transcriptome can provide valuable insights into tissue-specific gene expression patterns, dynamic changes in gene expression in response to varying environmental conditions, and regulatory mechanisms.

In conclusion, transcriptome-based gene prediction employs RNA-seq data to determine transcribed regions within a genome and infer gene boundaries. It is an effective method for characterising gene expression patterns, identifying novel genes, and identifying alternative splicing events. To substantiate predicted gene structures and expression patterns, however, experimental validation is crucial. Integration with other gene prediction methods can enhance gene annotations’ precision and comprehensiveness.

Comparative genomics

• Comparative genomics is an approach that compares the genomes of various species to identify similarities and differences in order to comprehend evolutionary relationships and functional components.
• Comparative genomics can be used to identify conserved regions of the genome that correspond presumably to genes in the context of gene prediction.
• The underlying assumption is that genes with similar functions across species have conserved sequences and structures.
• Depending on the research goals, comparative genomics can be applied to both closely related and distantly related species.
• The procedure involves aligning the genomic sequences of various species in order to identify sequence-conserved regions.
• It is probable that conserved regions contain protein-coding genes, regulatory elements, and other functional elements.
• Researchers can identify gene orthologs (genes in various species that share a common ancestor) and paralogs (genes resulting from gene duplication events) by comparing the genomes of different species.
• Comparative genomics can aid in the prediction of gene boundaries, exon-intron structures, and regulatory regions based on patterns of conservation observed in multiple species.
• Phylogenetic analysis can be used to infer the evolutionary relationships between genes and to trace the origin and evolution of genes across species.
• Comparative genomics can also assist in the identification of genes that are exclusive to a particular species or lineage, thereby shedding light on species-specific adaptations.
• Frequently, experimental validation, such as functional assays or gene expression studies, is required to corroborate the predicted gene functions and regulatory interactions.

In conclusion, comparative genomics is an effective method for gene prediction that is based on contrasting the genomic sequences of various species. It can aid in the identification of conserved regions and the inference of gene boundaries, exon-intron structures, and regulatory elements. Comparative genomics sheds light on gene evolution, functional elements, and species-specific adaptations by analysing the similarities and differences between genomes. To corroborate the predicted gene functions and regulatory interactions, experimental validation is essential.

Machine learning and deep learning

Machine learning and deep learning are computational approaches that involve training models to learn patterns and make predictions from data. Here’s some information about machine learning and deep learning presented in a point format:

Machine Learning:

• The subfield of artificial intelligence known as machine learning focuses on the development of algorithms and models that can learn from data and make predictions or decisions without being explicitly programmed.
• It entails training models on labelled training data, where both the input features and the output labels are known.
• Algorithms that employ machine learning discover patterns and relationships within training data and use them to make predictions or classify new, unseen data.
• Decision trees, random forests, support vector machines (SVM), k-nearest neighbours (KNN), and naive Bayes are typical machine learning algorithms.
• Feature engineering, the process of selecting and transforming pertinent data features, is a crucial phase in machine learning.
• In order to optimise model performance, machine learning algorithms must tune hyperparameters.
• Common machine learning paradigms include supervised learning, unsupervised learning, and reinforcement learning.
• Using labelled training data, supervised learning involves predicting output labels from input features.
• The objective of unsupervised learning is to discover patterns or structures in unlabeled data.
• The process of learning optimal actions based on environmental feedback is known as reinforcement learning.
• The efficacy of machine learning models is measured using evaluation metrics including accuracy, precision, recall, and F1-score.
• Several domains, such as image recognition, natural language processing, recommendation systems, and anomaly detection, make extensive use of machine learning.

Deep Learning:

• Deep learning is a subfield of machine learning that concentrates on training deep neural networks, which are models with multiple layers of artificial neurons that are interconnected.
• Deep neural networks are designed to discover hierarchical data representations, with each layer extracting increasingly intricate features.
• In a variety of tasks, deep learning models, such as convolutional neural networks (CNNs) for image data and recurrent neural networks (RNNs) for sequential data, have obtained remarkable success.
• For training, deep learning models require vast quantities of labelled data and frequently benefit from high-performance computing resources.
• Training deep neural networks requires optimisation algorithms such as stochastic gradient descent (SGD) and backpropagation, which adjust the network’s weights based on the difference between actual and predicted outputs.
• Deep learning models can autonomously discover pertinent features from unprocessed data, eliminating the need for manual feature engineering.
• In situations with limited labelled data, transfer learning, which utilises pre-trained deep learning models on large datasets, is prevalent.
• Deep learning has produced state-of-the-art results in numerous domains, including image classification, object detection, speech recognition, natural language processing, and many others.
• Metrics such as accuracy, precision, recall, and a variety of task-specific metrics are frequently used to evaluate deep learning models.
• Deep learning frameworks, such as TensorFlow, Keras, and PyTorch, offer tools and libraries for efficiently developing and training deep learning models.

In conclusion, machine learning and deep learning are effective techniques for extracting patterns from data and making predictions. Deep learning involves training deep neural networks to automatically learn hierarchical representations, whereas machine learning focuses on training models with labelled data. These techniques have revolutionised numerous disciplines and continue to advance artificial intelligence’s capabilities.

Gene Prediction Advantages

1. Genome Annotation: Gene prediction provides crucial information about the location, boundaries, and potential functions of genes within a genome, facilitating genome annotation and understanding of gene organization.
2. Functional Insights: Gene prediction helps identify protein-coding genes, non-coding RNA genes, and other functional elements, providing insights into gene functions, regulatory regions, and potential interactions.
3. Comparative Analysis: Gene prediction enables comparative genomics, allowing researchers to compare gene structures, gene families, and evolutionary relationships across different species, aiding in studying gene evolution and functional conservation.
4. High-throughput Analysis: Gene prediction algorithms enable automated analysis of large-scale genomic data, making it possible to process vast amounts of DNA sequence data efficiently.
5. Prediction of Novel Genes: Gene prediction methods can identify novel genes that may not have been previously annotated or discovered, expanding our knowledge of the genomic repertoire and potentially leading to the discovery of new functional elements.

Limitations of Gene Prediction

1. Prediction Errors: Gene prediction methods are not perfect and can make errors in identifying gene boundaries, exon-intron structures, and distinguishing coding from non-coding regions. Experimental validation is often required to confirm predicted gene structures.
2. Complex Gene Structures: Genes can have complex structures, such as alternative splicing, overlapping genes, or nested genes, which can pose challenges for accurate prediction using computational methods alone.
3. Genomic Variability: Gene prediction algorithms may not account for genomic variations, including insertions, deletions, or rearrangements, which can affect gene structure and expression. This can lead to inaccuracies in gene prediction, particularly in highly variable regions of the genome.
4. Limited Training Data: Some gene prediction methods rely on training data from known genes in related species. Lack of sufficient training data or divergent genomic features can impact the accuracy of predictions, particularly for poorly characterized or evolutionarily distant genomes.
5. Non-Coding Region Identification: Identifying non-coding regions, such as regulatory elements or functional RNA genes, can be challenging as they often lack well-defined sequence motifs and can exhibit low conservation across species.
6. Integration of Multiple Data Sources: Gene prediction methods often benefit from integrating multiple data sources, such as transcriptome data or homology information. However, data integration can be complex and may introduce additional challenges in terms of data quality, compatibility, and interpretation.

Applications of Gene Prediction

Gene prediction plays a crucial role in the field of genomics and is crucial to numerous aspects of biological research. Here are several explanations why gene prediction is so crucial:

• Understanding Genome Structure: Gene prediction facilitates the identification and annotation of the locations and structures of genes within a genome. It enables researchers to determine the beginning and end points of genes, as well as their exon-intron boundaries and other crucial characteristics. Understanding the organisation and structure of genomes requires this information.
• Functional Annotation: Gene prediction is indispensable for functional annotation, which entails designating biological functions to genes. To infer the functions of predicted genes, they can be compared to known genes in databases. This knowledge aids in the comprehension of the molecular mechanisms underlying biological processes such as development, disease, and evolution.
• Comparative Genomics: Gene prediction permits gene comparisons between species. Researchers can identify conserved genes and analyse their functions, evolutionary history, and relationships by predicting genes in related organisms. Comparative genomics offers insights into the evolution of genes and genomes and facilitates the comprehension of the genetic basis for species differences.
• Drug Discovery and Biotechnology: Gene prediction is essential for identifying potential drug targets in Drug Discovery and Biotechnology. Targeting genes predicted to be implicated in disease processes or specific pathways enables drug development. In addition, gene prediction facilitates the identification of desirable genes in agricultural and industrial biotechnology. This information contributes to the development of genetically modified organisms with enhanced properties, such as increased crop yields or enhanced production of valuable compounds.
• Disease Research: Accurate gene prediction is indispensable for investigating the genetic basis of disease. It makes it possible to identify disease-associated genes and genetic variants that contribute to disease susceptibility. Understanding the genetic components of diseases assists with disease diagnosis, treatment, and prevention.
• Genome Annotation and Database Construction: Gene prediction is the first stage in genome annotation, in which genomic features and functional elements are identified and annotated. Accurate gene predictions contribute to the development of exhaustive and trustworthy genome databases, which are invaluable scientific resources.
• Evolutionary Studies: Gene prediction facilitates the study of the evolutionary history of genes and genomes. By comparing gene predictions across various species, scientists can infer ancestral genes and trace their evolutionary trajectories. This information aids in comprehending the origins and diversification of genes, as well as the mechanisms that drive the evolution of the genome.

FAQ

References

1. Burge, C.B. and Karlin, S. (1997). Prediction of complete gene structures in human genomic DNA. Journal of Molecular Biology, 268(1), 78-94.
2. Majoros, W.H., Pertea, M., and Salzberg, S.L. (2004). TigrScan and GlimmerHMM: Two open source ab initio eukaryotic gene-finders. Bioinformatics, 20(16), 2878-2879.
3. Allen, J.E., Pertea, M., and Salzberg, S.L. (2004). Computational gene prediction using multiple sources of evidence. Genome Research, 14(1), 142-148.
4. Stanke, M., Schöffmann, O., Morgenstern, B., and Waack, S. (2006). Gene prediction in eukaryotes with a generalized hidden Markov model that uses hints from external sources. BMC Bioinformatics, 7(1), 62.
5. Lomsadze, A., Ter-Hovhannisyan, V., Chernoff, Y.O., and Borodovsky, M. (2005). Gene identification in novel eukaryotic genomes by self-training algorithm. Nucleic Acids Research, 33(20), 6494-6506.
6. Lowe, T.M., and Eddy, S.R. (1997). tRNAscan-SE: A program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Research, 25(5), 955-964.
7. Korf, I. (2004). Gene finding in novel genomes. BMC Bioinformatics, 5(1), 59.
8. Camacho, C., Coulouris, G., Avagyan, V., Ma, N., Papadopoulos, J., Bealer, K., and Madden, T.L. (2009). BLAST+: Architecture and applications. BMC Bioinformatics, 10(1), 421.
9. Allen, J.E., and Salzberg, S.L. (2005). Comparative genomics: Genome-wide analysis in metazoan eukaryotes. Nature Reviews Genetics, 6(12), 917-927.
10. Bateman, A., Martin, M.J., O’Donovan, C., Magrane, M., Alpi, E., Antunes, R., … and Leinonen, R. (2017). UniProt: The universal protein knowledgebase. Nucleic Acids Research, 45(D1), D158-D169.