1 The field of Genome annotation pays a lot of attentions where the ability to collect and analysis genomical data can provide strong indicator for the study of life\cite{Eisen2007}. Four of genome annotation centers, (such as, \textit{NCBI\cite{Sayers01012011}, Dogma \cite{RDogma}, cpBase \cite{de2002comparative}, CpGAVAS \cite{liu2012cpgavas}, and CEGMA\cite{parra2007cegma}}), present various types of annotations tools (i.e cost-effective sequencing methods\cite{Bakke2009}) on different annotation levels. Generally, one of three methods of gene finding in annotated genome can be categorized using these centers: \textit{alignment-based, composition based, or combination of both\cite{parra2007cegma}}. The alignment-based method is used when we try to predict a coding gene (i.e. genes that produce proteins) by aligning DNA sequence of gene to the protein of cDNA sequence of homology\cite{parra2007cegma}. This approache also is used in GeneWise\cite{birney2004genewise}. Composition-based mothod (known as \textit{ab initio}) is based on a probabilistic model of gene structure to find genes and/or new genes according to the probability gene value (GeneID\cite{parra2000geneid}). In this section, we will consider a new method of finding core genes from large amount of chloroplast genomes, as a solution of the problem resulting from the method stated in section two. This method is based on extracting gene features. A general overview of the system is illustrated in Figure \ref{Fig1}.\\
5 \includegraphics[width=0.7\textwidth]{generalView}
6 \caption{A general overview of the system}\label{Fig1}
9 In Figure 1, we illustrate the general overview of system pipeline: \textit{Database, Genomes annotation, Core extraction,} and \textit{relationships}. We will give a short discussion for each stage of the model in order to understand all core extraction process. This work starts with a gene Bank database; however, many international Banks for nucleotide sequence databases (such as, \textit{GenBank} \cite{Sayers01012011} in USA, \textit{EMBL-Bank} \cite{apweiler1985swiss} in Europe, and \textit{DDBJ} \cite{sugawara2008ddbj} in Japon) where exist to store various genomes and DNA species. Different Biological tools are provided to analyse and annotate genomes by interacting with these databases to align and extract sequences to predict genes. The database in this model must be taken from any confident data source that store annotated and/or unannotated chloroplast genomes. We will consider GenBank-NCBI \cite{Sayers01012011} database to be our nucleotide sequences database. Annotation (as the second stage) is considered to be the first important task for Extract Gene Features. Good annotation tool lead us to extracts good gene feature. In this paper, two annotation techniques from \textit{NCBI, and Dogma} are used to extract \textit{genes features}. Extracting Gene feature (as a third stage) can be anything like (genes names, gene sequences, protein sequence,...etc). Our methodologies in this paper consider gene names, genes counts, and gene sequences for extracting core genes and producing chloroplast evolutionary tree. \\
10 In last stage, features visualization represents methods to visualize genomes and/or gene evolution in chloroplast. By using the forms of (tables, phylogenetic trees, graphs,...,etc) to organize and represent genomes relationships can achieve the goal of gene evolution with what the biological expert needs. In addition, compare these forms with another annotation tool forms for large population of chloroplast genomes give us biological perspective to the nature of chloroplasts evolution. \\
11 A Local database attached with each pipe stage is used to store all the informations of extraction process. The output from each stage in our system will be an input to the second stage and so on.
13 \subsection{Genomes Samples}
14 In this research, we retrieved genomes of Chloroplasts from NCBI. Ninety nine genome of them were considered to work with. These genomes lies in the eleven type of chloroplast families. The distribution of genomes is illustrated in detail in Table~\ref{Tab2}.
16 \input{population_Table}
18 \subsection{Genome Annotation Techniques}
19 Genome annotation is the second stage in the model pipeline. Many techniques were developed to annotate chloroplast genomes but the problem is that they vary in the number and type of predicting genes (i.e the ability to predict genes and \textit{for example: Transfer RNA (tRNA)} and \textit{Ribosomal RNA (rRNA)} genes). Two annotation techniques from NCBI and Dogma are considered to analyse chloroplast genomes to examine the accuracy of predicted coding genes.
21 \subsubsection{genome annotation from NCBI}
22 The objective from this step is to organize genes, solve gene duplications, and generate sets of genes from each genome. The input to the system is our list of chloroplast genomes, annotated from NCBI. All genomes stored as \textit{.fasta} files which have a collection of protein coding genes\cite{parra2007cegma,RDogma}(gene that produce proteins) with its coding sequences.
23 As a preprocessing step to achieve the set of core genes, we need to analyse these genomes (using \textit{BioPython} package\cite{chapman2000biopython}), to extracting all information needed to find the core genes. The process starts by converting each genome from fasta format to GenVision\cite{geneVision} formats from DNASTAR. The outputs from this operation are lists of genes for each genome, their genes names and gene counts. In this stage, we accumulate some Gene duplications for each treated genome. In other words, duplication in gene name can comes from genes fragments, (e.g. gene fragments treated with NCBI), as long as chloroplast DNA sequences. To ensure that all the duplications are removed, each list of genes is translated into a set of genes. NCBI genome annotation produce genes except \textit{Ribosomal rRNA}.
25 \subsubsection{Genome annotation from Dogma}
26 Dogma is an annotation tool developed in the university of Texas in 2004. Dogma is an abbreviation of (\textit{Dual Organellar GenoMe Annotator}) for plant chloroplast and animal mitochondrial genomes.
27 It has its own database for translating the genome in all six reading frames and query the amino acid sequence database using Blast\cite{altschul1990basic}(i.e Blastx) with various parameters. Further more, identify protein coding genes in the input genome based on sequence similarity of genes in Dogma database. In addition, it can produce the \textit{Transfer RNAs (tRNA)}, and the \textit{Ribosomal RNAs (rRNA)} and verifies their start and end positions rather than NCBI annotation tool. There are no gene duplication with dogma after solving gene fragmentation. \\
28 Genome Annotation with dogma can be the key difference of extracting core genes. The step of annotation divided into two tasks: First, It starts to annotate complete chloroplast genome (i.e \textit{Unannotate genome from NCBI} by using Dogma web tool. This process was done manually. The output from dogma is considered to be collection of coding genes file for each genome in the form of GeneVision file format.
29 Where the second task is to solve gene fragments. Two methods used to solve genes duplication for extract core genes. First, for the method based on gene name, all the duplications are removed, where each list of genes is translated into a set of genes. Second, for the method of gene quality test, defragment process starts immediately to solve fragments of coding genes for each genome to avoid gene duplication. In each iteration, this process starts by taking one gene from gene list, search for gene duplication, if exists, look on the orientation of the fragment sequence: if it is positive, then appending fragment sequence to gene file. Otherwise, the process applies reverse complement operations on gene sequence and append it to gene file. Additional process applied to check start and stop codon and try to find appropriate start and end codon in case of missing. All genomes after this stage are fully annotated, their genes were de-fragmented, genes lists and counts were identified.\\
31 \subsection{Core Genes Extraction}
32 The goal of this step is to extract maximum core genes from sets of genes. The methodology of finding core genes is as follow: \\
34 \subsubsection{Pre-Processing}
35 We apply two pre-processing methods for organize and prepare genomes data, one method based on gene name and count, and the second method is based on sequence quality control test.\\
36 In the first method, preparing chloroplasts genomes to extract core genes based on gene name and count starts after annotation process because genomes vary in genes counts and types according to the annotation used method. Then we store each genome in the database under genome name with the set of genes names. Genes counts can extracted simply by a specific length command. \textit{Intersection core matrix} will apply then to extract the core genes. The problem with this method is how we can quarantine that the gene predicted in core genes is the same gene in leaf genomes?. To answer this question, if the sequence of any gene in a genome annotated from dogma and NCBI are similar with respect to a threshold, we do not have any problem with this method. Otherwise, we have a problem, because we can not decide which sequence goes to a gene in core genes.
37 The second pre-processing method state: we can predict the best annotated genome by merge the annotated genomes from NCBI and dogma based on the quality of genes names and sequences test. To generate all quality genes of each genome. the hypothesis state: Any gene will be in predicted genome if and only if the annotated genes between NCBI and Dogma pass a specific threshold of\textit{quality control test}. To accept the quality test, we applied Needle-man Wunch algorithm to compare two gene sequences with respect to pass a threshold. If the alignment score pass this threshold, then the gene will be in the predicted genome. Otherwise, the gene will be ignored. After predicting all genomes, \textit{Intersection core matrix} will apply on these new genomes to extract core genes. As shown in Algorithm \ref{Alg3:thirdM}.
40 \caption{Extract new genome based on Gene Quality test}
43 \REQUIRE $Gname \leftarrow \text{Genome Name}, Threshold \leftarrow 65$
44 \ENSURE $geneList \leftarrow \text{Quality genes}$
45 \STATE $dir(NCBI\_Genes) \leftarrow \text{NCBI genes of Gname}$
46 \STATE $dir(Dogma\_Genes) \leftarrow \text{Dogma genes of Gname}$
47 \STATE $geneList=\text{empty list}$
48 \STATE $common=set(dir(NCBI\_Genes)) \cap set(dir(Dogma\_Genes))$
49 \FOR{$\text{gene in common}$}
50 \STATE $g1 \leftarrow open(NCBI\_Genes(gene)).read()$
51 \STATE $g2 \leftarrow open(Dogma\_Genes(gene)).read()$
52 \STATE $score \leftarrow geneChk(g1,g2)$
53 \IF {$score > Threshold$}
54 \STATE $geneList \leftarrow gene$
61 \textbf{geneChk} is a subroutine, it is used to find the best similarity score between two gene sequences after applying operations like \textit{reverse, complement, and reverse complement}. The algorithm of geneChk is illustrated in Algorithm \ref{Alg3:genechk}.
64 \caption{Find the Maximum similarity score between two sequences}
67 \REQUIRE $gen1,gen2 \leftarrow \text{NCBI gene sequence, Dogma gene sequence}$
68 \ENSURE $\text{Maximum similarity score}$
69 \STATE $Score1 \leftarrow needle(gen1,gen2)$
70 \STATE $Score2 \leftarrow needle(gen1,Reverse(gen2))$
71 \STATE $Score3 \leftarrow needle(gen1,Complement(gen2))$
72 \STATE $Score4 \leftarrow needle(gen1,Reverse(Complement(gen2)))$
73 \IF {$max(Score1, Score2, Score3, Score4)==Score1$}
75 \ELSIF {$max(Score1, Score2, Score3, Score4)==Score2$}
77 \ELSIF {$max(Score1, Score2, Score3, Score4)==Score3$}
79 \ELSIF {$max(Score1, Score2, Score3, Score4)==Score4$}
85 \subsubsection{Intersection Core Matrix (\textit{ICM})}
87 The idea behind extracting core genes is to collect iteratively the maximum number of common genes between two genomes. To do so, the system builds an \textit{Intersection core matrix (ICM)}. ICM is a two dimensional symmetric matrix where each row and column represent one genome. Each position in ICM stores the \textit{intersection scores(IS)}. The Intersection Score is the cardinality number of a core genes which comes from intersecting one genome with other ones. Maximum cardinality results to select two genomes with their maximum core. Mathematically speaking, if we have an $n \times n$ matrix where $n=\text{number of genomes in local database}$, then lets consider:\\
90 Score=\max_{i<j}\vert x_i \cap x_j\vert
94 Where $x_i, x_j$ are elements in the matrix. The generation of a new core genes is depending on the cardinality value of intersection scores, we call it \textit{Score}:
95 $$\text{New Core} = \begin{cases}
96 \text{Ignored} & \text{if $\textit{Score}=0$;} \\
97 \text{new Core id} & \text{if $\textit{Score}>0$.}
100 if $\textit{Score}=0$ then we have \textit{disjoint relation} (i.e no common genes between two genomes). In this case the system ignores the genome that annul the core genes size. Otherwise, The system will removes these two genomes from ICM and add new core genomes with a \textit{coreID} of them to ICM for the calculation in next iteration. This process will reduce the size of ICM and repeat until all genomes are treated (i.e ICM has no more genomes).
101 We observe that ICM is very large because of the amount of data that it stores. This results to be time and memory consuming for calculating the intersection scores. To increase the speed of calculations, it is sufficient to only calculate the upper triangle scores. The time complexity for this process after enhancement is thus $O(\frac{(n-1).n}{2})$. Algorithm \ref{Alg1:ICM} illustrates the construction of the ICM matrix and the extraction of the core genes where \textit{GenomeList}, represents the database where all genomes data are stored. At each iteration, it computes the maximum core genes with its two genomes parents.
104 \caption{Extract Maximum Intersection Score}
107 \REQUIRE $L \leftarrow \text{genomes vectors}$
108 \ENSURE $B1 \leftarrow Max core vector$
109 \FOR{$i \leftarrow 0:len(L)-1$}
110 \STATE $core1 \leftarrow set(GenomeList[L[i]])$
111 \STATE $score1 \leftarrow 0$
112 \STATE $g1,g2 \leftarrow$ " "
113 \FOR{$j \leftarrow i+1:len(L)$}
114 \STATE $core2 \leftarrow set(GenomeList[L[i]])$
116 \STATE $Core \leftarrow core1 \cap core2$
117 \IF{$len(Core) > score1$}
118 \STATE $g1 \leftarrow L[i]$
119 \STATE $g2 \leftarrow L[j]$
120 \STATE $Score \leftarrow len(Core)$
121 \ELSIF{$len(Core) == 0$}
122 \STATE $g1 \leftarrow L[i]$
123 \STATE $g2 \leftarrow L[j]$
124 \STATE $Score \leftarrow -1$
128 \STATE $B1[score1] \leftarrow (g1,g2)$
134 \subsection{Features Visualization}
135 The goal here is to visualize the results by building a tree of evolution. All Core genes generated with their genes are very important information in the tree, because they can be viewed as an ancestor information for two genomes or more. Further more, each node represents a genome or core as \textit{(Genes count:Family name, Scientific names, Accession number)}, Edges represent numbers of lost genes from genomes-core or core-core relationship. The number of lost genes here can represent an important factor for evolution, it represents how much lost of genes for the species in same or different families. By the principle of classification, small number of gene lost among species indicate that those species are close to each other and belong to same family, while big genes lost means that we have an evolutionary relationship between species from different families. To see the picture clearly, Phylogenetic tree is an evolutionary tree generated also by the system. Generating this tree is based on the distances among genes sequences. There are many resources to build such tree (for example: PHYML\cite{guindon2005phyml}, RAxML{\cite{stamatakis2008raxml,stamatakis2005raxml}, BioNJ , and TNT\cite{goloboff2008tnt}}. We consider to use RAxML\cite{stamatakis2008raxml,stamatakis2005raxml} because it is very fast for build large trees even for hundered sequences, it is also accurate by calculating bootstrap.
139 \includegraphics[width=0.7\textwidth]{Whole_system}
140 \caption{Total overview of the system pipeline}\label{wholesystem}
143 \section{Implementation}
144 We implemented four algorithms to extract maximum core genes from large amount of chloroplast genomes. Two algorithms used to extract core genes based on NCBI annotation, and the others based on dogma annotation tool. Evolutionary tree generated as a result from each method implementation. In this section, we will present the four methods, and how they can extract maximum core genes?, and how the developed code will generate the evolutionary tree.
146 \subsection{Extract Core Genes based on Gene Contents}
148 \subsubsection{Core Genes based on NCBI Annotation}
149 The first idea to construct the core genome is based on the extraction of Genes names (as gene presence or absence). For instant, in this stage neither sequence comparison nor new annotation were made, we just want to extract all genes with counts stored in each chloroplast genome, then find the intersection core genes based on gene names. \\
150 The pipeline of extracting core genes can summarize in the following steps:\\
151 First, we apply the genome annotation method using NCBI annotation tool. Genome quality check can be used in this step to ensure that genomes pass some quality condition. Then, the system lunch annotation process using NCBI to extract code genes (i.e \textit{exons}) and solve gene fragments. From NCBI, we did not observe any problem with genes fragments, but there are a problem of genes orthography (e.g two different genes sequences with same gene name). After we obtain all annotated genomes from NCBI to the local database, the code will then automatically will generate GenVision\cite{geneVision} file format to lunch the second step to extract coding genes names and counts. The competition will start by building intersection matrix to intersect genomes vectors in the local database with the others. New core vector for two leaf vectors will generate and a specific \textit{CoreId} will assign to it. an evolutionary tree will take place by using all data generated from step 1 and 2. The tree will also display the amount of genes lost from each intersection iteration. A specific excel file will be generated that store all the data in local database. The whole operation illstrate in Figure \ref{NCBI:geneextraction}.
154 \subsubsection{Core Genes based on Dogma Annotation}
155 The main goal is to get as much as possible the core genes of maximum coding genes names. According to NCBI annotation problem based on \cite{Bakke2009}, annotation method like dogma can give us more reliable coding genes than NCBI. This is because NCBI annotation can carry some annotation and gene identification errors. The general overview of whole process of extraction illustrated in figure \ref{wholesystem}.
157 \subsubsection{extracting core genes based on genes names and counts}
159 extracting core genes based on genes names and counts summarized in the following steps:\\
161 \item We apply the genome annotation manually using Dogma annotation tool.
162 \item Analysing genomes to store lists of code genes names (i.e \textit{exons}). solve gene fragments is done by using first method in solve gene fragments. The output from annotation process with dogma is genomes files in GenVision file format. Sets of genes were stored in the database.
163 \item Generate ICM matrix to calculate maximum core genes.
164 \item Draw the evolutionary tree by extracted all genes sequences from each core. Then applying multiple alignment process on the sequences to calculate the distance among cores to draw a phylogenetic tree.
173 the code will lunch genes de-fragments process to avoid genes duplications. little problems of genes orthography (e.g two different genes sequences with same gene name) where exists. After we obtain all annotated genomes from dogma, we store it in the local database. The code will then automatically lunch the second step to extract coding genes names and counts. The competition will start by building intersection matrix to intersect genomes vectors in the local database with the others. New core vector for two leaf vectors will generate and a specific \textit{CoreId} will assign to it. an evolutionary tree will take place by using all data generated from step 1 and 2. The tree will also display the amount of genes lost from each intersection iteration. A specific excel file will be generated that store all the data in local database. The whole operation illustrate in Figure \ref{dogma:geneextraction}.
178 \includegraphics[width=0.7\textwidth]{Dogma_geneextraction}
179 \caption{Extract core genes based on Dogma gene names and counts}\label{dogma:geneextraction}
182 The main drawback from the method of extracting core genes based on gene names and counts is that we can not depending only on genes names because of three causes: first, the genome may have not totally named (This can be found in early versions of NCBI genomes), so we will have some lost sequences. Second, we may have two genes sharing the same name, while their sequences are different. Third, we need to annotate all the genomes.
184 \subsection{Extract Core Genes based on Gene Quality Control}
185 The main idea from this method is to focus on genes quality to predict maximum core genes. By comparing only genes names or genes sequences from one annotation tool is not enough. The question here, does the predicted gene from NCBI is the same gene predicted by Dogma based on gene name and gene sequence?. If yes, then we can predict new quality genomes based on quality control test with a specific threshold. Predicted Genomes comes from merging two annotation techniques. While if no, we can not depending neither on NCBI nor Dogma because of annotation error. Core genes can by predicted by using one of the previous methods.
187 This method summarized in the following steps:\\
190 \item Retrieve the annotation of all genomes from NCBI and Dogma: in this step, we apply the annotation of all chloroplast genomes in the database using NCBI annotation and Dogma annotation tool.
191 \item Predict quality genomes: the process is to pick a genome annotation from two techniques, extracting all common genes based on genes names, then applying Needle-man wunch algorithm to align the two sequences based on a specific threshold. If the alignment score pass the threshold, then this gene will removed from the competition and store it in quality genome by saving its name with the largest gene sequence with respect to start and end codons. All quality genomes will store in the form of GenVision file format.
192 \item Extract Core genes: from the above two steps, we will have new genomes with quality genes, ofcourse, we have some genes lost here, because dogma produced tRNA and rRNA genes while NCBI did not generate them and vise-versa. Build ICM to extract core genes will be sufficient because we already check their sequences.
193 \item Display tree: An evolution tree then will be display based on the intersections of quality genomes.
