2 These last years the cost of sequencing genomes has been greatly
3 reduced, and thus more and more genomes are sequenced. Therefore
4 automatic annotation tools are required to deal with this continuously
5 increasing amount of genomical data. Moreover, a reliable and accurate
6 genome annotation process is needed in order to provide strong
7 indicators for the study of life\cite{Eisen2007}.
9 Various annotation tools (\emph{i.e.}, cost-effective sequencing
10 methods\cite{Bakke2009}) producing genomic annotations at many levels
11 of detail have been designed by different annotation centers. Among
12 the major annotation centers we can notice NCBI\cite{Sayers01012011},
13 Dogma \cite{RDogma}, cpBase \cite{de2002comparative},
14 CpGAVAS \cite{liu2012cpgavas}, and
15 CEGMA\cite{parra2007cegma}. Usually, previous studies used one out of
16 three methods for finding genes in annoted genomes using data from
17 these centers: \textit{alignment-based}, \textit{composition based},
18 or a combination of both~\cite{parra2007cegma}. The alignment-based
19 method is used when trying to predict a coding gene (\emph{i.e.}.
20 genes that produce proteins) by aligning a genomic DNA sequence with a
21 cDNA sequence coding an homologous protein \cite{parra2007cegma}.
22 This approach is also used in GeneWise\cite{birney2004genewise}. The
23 alternative method, the composition-based one (also known
24 as \textit{ab initio}) is based on a probabilistic model of gene
25 structure to find genes according to the gene value probability
26 (GeneID \cite{parra2000geneid}). Such annotated genomic data will be
27 used to overcome the limitation of the first method described in the
28 previous section. In fact, the second method we propose finds core
29 genes from large amount of chloroplast genomes through genomic
32 Figure~\ref{Fig1} presents an overview of the entire method pipeline.
33 More precisely, the second method consists of three
34 stages: \textit{Genome annotation}, \textit{Core extraction},
35 and \textit{Features Visualization} which highlights the
36 relationships. To understand the whole core extraction process, we
37 describe briefly each stage below. More details will be given in the
38 coming subsections. The method uses as starting point some sequence
39 database chosen among the many international databases storing
40 nucleotide sequences, like the GenBank at NBCI \cite{Sayers01012011},
41 the \textit{EMBL-Bank} \cite{apweiler1985swiss} in Europe
42 or \textit{DDBJ} \cite{sugawara2008ddbj} in Japan. Different
43 biological tools can analyze and annotate genomes by interacting with
44 these databases to align and extract sequences to predict genes. The
45 database in our method must be taken from any confident data source
46 that stores annotated and/or unannotated chloroplast genomes. We have
47 considered the GenBank-NCBI \cite{Sayers01012011} database as sequence
48 database: 99~genomes of chloroplasts were retrieved. These genomes
49 lie in the eleven type of chloroplast families and Table \ref{Tab2}
50 summarizes their distribution in our dataset.
54 \includegraphics[width=0.75\textwidth]{generalView}
55 \caption{A general overview of the annotation-based approach}\label{Fig1}
58 Annotation, which is the first stage, is an important task for
59 extracting gene features. Indeed, to extract good gene feature, a good
60 annotation tool is obviously required. The extraction of gene feature, the next stage, can be anything like gene names, gene sequences, protein sequences, and so on. Our method considers gene names, gene counts, and gene sequence for extracting core genes and producing chloroplast evolutionary tree. The final stage allows to visualize genomes and/or gene evolution in chloroplast. Therefore we use representations like tables, phylogenetic trees, graphs, etc. to organize and show genomes relationships, and thus achieve the goal of representing gene
61 evolution. In addition, comparing these representations with ones issued from another annotation tool dedicated to large population of chloroplast genomes give us biological perspectives to the nature of chloroplasts evolution. Notice that a local database linked with each pipe stage is used to store all the informations produced during the process.
63 \input{population_Table}
65 \subsection{Genome annotation techniques}
67 To obtain relevant annotated genomes, two annotation techniques from NCBI and Dogma are used. For the first stage, genome annotation, many techniques have been developed to annotate chloroplast genomes. These techniques differ
68 from each others in the number and type of predicted genes (for example: \textit{Transfer RNA (tRNA)} and \textit{Ribosomal RNA (rRNA)} genes). Two annotation techniques from NCBI and Dogma are considered to analyze chloroplast genomes.
70 \subsubsection{Genome annotation from NCBI}
72 The objective is to generate sets of genes from each genome so that
73 genes are organized without any duplication. The input is a list of
74 chloroplast genomes annotated from NCBI. More precisely, all genomes
75 are stored as \textit{.fasta} files which consists in a collection of
76 protein coding genes\cite{parra2007cegma,RDogma} (gene that produce
77 proteins) organized in coding sequences. To be able build the set of
78 core genes, we need to preprocess these genomes
79 using \textit{BioPython} package \cite{chapman2000biopython}. This
80 step starts by converting each genome from FASTA file format to
81 GenVision \cite{geneVision} format from DNASTAR. Each genome is thus
82 converted in a list of genes, with gene names and gene counts. Gene
83 name duplications can be accumulated during the treatment of a genome.
84 These duplications come from gene fragments (\emph{e.g.} gene
85 fragments treated with NCBI) and from chloroplast DNA sequences. To
86 ensure that all the duplications are removed, each list of gene is
87 translated into a set of genes. Note that NCBI genome annotation
88 produces genes except \textit{Ribosomal (rRNA)} genes.
90 \subsubsection{Genome annotation from Dogma}
92 Dogma stands for \textit{Dual Organellar GenoMe Annotator}. It is an
93 annotation tool developed at University of Texas in 2004 for plant
94 chloroplast and animal mitochondrial genomes. This tool has its own
95 database for translating a genome in all six reading frames and
96 queries the amino acid sequence database using
97 BLAST \cite{altschul1990basic} (\emph{i.e.} Blastx) with various
98 parameters. Protein coding genes are identified in an input genome
99 using sequence similarity of genes in Dogma database. In addition in
100 comparison with NCBI annotation tool, Dogma can produce
101 both \textit{Transfer RNAs (tRNA)} and \textit{Ribosomal RNAs (rRNA)},
102 verify their start and end positions. further more, there is no gene duplication with gene annotations from Dogma after applying gene de-fragmentation process. In fact, genome annotation with Dogma can be the key difference when extracting core genes.
104 The Dogma annotation process is divided into two tasks. First, we
105 manually annotate chloroplast genomes using Dogma web tool. The output
106 of this step is supposed to be a collection of coding genes files for
107 each genome, organized in GeneVision file. The second task is to solve
108 the gene duplication problem and therefore we have used two
109 methods. The first method, based on gene name, translates each genome
110 into a set of genes without duplicates. The second method avoid gene
111 duplication through a defragment process. In each iteration, this
112 process starts by taking a gene from gene list, searches for gene
113 duplication, if a duplication is found, it looks on the orientation of
114 the fragment sequence. If it is positive it appends directly the
115 sequence to gene files. Otherwise reverse complement operations are
116 applied on the sequence, which is then also append to gene files.
117 Finally, a check for missing start and stop codons is performed. At
118 the end of the annotation process, all the genomes are fully
119 annotated, their genes are defragmented, and gene counts are
122 \subsection{Core genes extraction}
124 The goal of this stage is to extract maximum core genes from sets of
125 genes. To find core genes, the following methodology is applied.
127 \subsubsection{Preprocessing}
129 In order to extract core genomes in a suitable manner, the genomic
130 data are preprocessed with two methods: on the one hand a method based
131 on gene name and count, and on the other hand a method based on a
132 sequence quality control test.
134 In the first method, we extract a list of genes from each chloroplast
135 genome. Then we store this list of genes in the database under genome
136 nam and genes counts can be extracted by a specific length command.
137 The \textit{Intersection Core Matrix}, described in next subsection,
138 is then computed to extract the core genes. The problem with this
139 method can be stated as follows: how can we ensure that the gene which
140 is predicted in core genes is the same gene in leaf genomes? The
141 answer to this problem is that if the sequences of any gene in a
142 genome annotated from Dogma and NCBI are similar with respect to a
143 given threshold, the method is operational when the sequences are not similar. The problem of attribution of a sequence to a gene in the core genome come to light.
145 The second method is based on the underlying idea that it is possible to predict the the best annotated genome by merging the annotated genomes from NCBI
146 and Dogma according to a quality test on genes names and sequences. To
147 obtain all quality genes of each genome, we consider the following
148 hypothesis: any gene will appear in the predicted genome if and only
149 if the annotated genes in NCBI and Dogma pass a specific threshold
150 of \textit{quality control test}. In fact, the Needle-man Wunch
151 algorithm is applied to compare both sequences with respect to a
152 threshold. If the alignment score is above the threshold, then the
153 gene will be retained in the predicted genome, otherwise the gene is
154 ignored. Once the prediction of all genomes is done,
155 the \textit{Intersection Core Matrix} is computed on these new genomes
156 to extract core genes, as explained in Algorithm \ref{Alg3:thirdM}.
159 \caption{Extract new genome based on gene quality test}
162 \REQUIRE $Gname \leftarrow \text{Genome Name}, Threshold \leftarrow 65$
163 \ENSURE $geneList \leftarrow \text{Quality genes}$
164 \STATE $dir(NCBI\_Genes) \leftarrow \text{NCBI genes of Gname}$
165 \STATE $dir(Dogma\_Genes) \leftarrow \text{Dogma genes of Gname}$
166 \STATE $geneList=\text{empty list}$
167 \STATE $common=set(dir(NCBI\_Genes)) \cap set(dir(Dogma\_Genes))$
168 \FOR{$\text{gene in common}$}
169 \STATE $g1 \leftarrow open(NCBI\_Genes(gene)).read()$
170 \STATE $g2 \leftarrow open(Dogma\_Genes(gene)).read()$
171 \STATE $score \leftarrow geneChk(g1,g2)$
172 \IF {$score > Threshold$}
173 \STATE $geneList \leftarrow gene$
180 \textbf{geneChk} is a subroutine used to find the best similarity score between
181 two gene sequences after applying operations like \textit{reverse}, {\it complement},
182 and {\it reverse complement}. Algorithm~\ref{Alg3:genechk} gives the outline of
186 \caption{Find the Maximum Similarity Score between two sequences}
189 \REQUIRE $g1,g2 \leftarrow \text{NCBI gene sequence, Dogma gene sequence}$
190 \ENSURE $\text{Maximum similarity score}$
191 \STATE $score1 \leftarrow needle(g1,g2)$
192 \STATE $score2 \leftarrow needle(g1,Reverse(g2))$
193 \STATE $score3 \leftarrow needle(g1,Complement(g2))$
194 \STATE $score4 \leftarrow needle(g1,Reverse(Complement(g2)))$
195 \RETURN $max(score1,score2,score3,score4)$
199 \subsubsection{Intersection Core Matrix (\textit{ICM})}
201 To extract core genes, we iteratively collect the maximum number of
202 common genes between genomes and therefore during this stage
203 an \textit{Intersection Core Matrix} (ICM) is built. ICM is a two
204 dimensional symmetric matrix where each row and each column correspond
205 to one genome. Hence, an element of the matrix stores
206 the \textit{Intersection Score} (IS): the cardinality of the core
207 genes set obtained by intersecting one genome with another
208 one. Maximum cardinality results in selecting the two genomes having
209 the maximum score. Mathematically speaking, if we have $n$ genomes in
210 local database, the ICM is an $n \times n$ matrix whose elements
213 score_{ij}=\vert g_i \cap g_j\vert
216 \noindent where $1 \leq i \leq n$, $1 \leq j \leq n$, and $g_i, g_j$ are
217 genomes. The generation of a new core gene depends obviously on the
218 value of the intersection scores $score_{ij}$. More precisely, the
219 idea is to consider a pair of genomes such that their score is the
220 largest element in ICM. These two genomes are then removed from matrix
221 and the resulting new core genome is added for the next iteration.
222 The ICM is then updated to take into account the new core gene: new IS
223 values are computed for it. This process is repeated until no new core
224 gene can be obtained.
226 We can observe that the ICM is very large due to the amount of
227 data. As a consequence, the computation of the intersection scores is
228 both time and memory consuming. However, since ICM is a symetric
229 matrix we can reduce the computation overhead by considering only its
230 triangular upper part. The time complexity for this process after
231 enhancement is thus $O(\frac{n.(n-1)}{2})$. Algorithm ~\ref{Alg1:ICM}
232 illustrates the construction of the ICM matrix and the extraction of
233 the core genes, where \textit{GenomeList} represents the database
234 storing all genomes data. At each iteration, it computes the maximum
235 core genes with its two genomes parents.
237 % ALGORITHM HAS BEEN REWRITTEN
240 \caption{Extract Maximum Intersection Score}
243 \REQUIRE $L \leftarrow \text{genomes sets}$
244 \ENSURE $B1 \leftarrow \text{Max Core set}$
245 \FOR{$i \leftarrow 1:len(L)-1$}
246 \STATE $score \leftarrow 0$
247 \STATE $core1 \leftarrow set(GenomeList[L[i]])$
248 \STATE $g1 \leftarrow L[i]$
249 \FOR{$j \leftarrow i+1:len(L)$}
250 \STATE $core2 \leftarrow set(GenomeList[L[j]])$
251 \STATE $core \leftarrow core1 \cap core2$
252 \IF{$len(core) > score$}
253 \STATE $score \leftarrow len(core)$
254 \STATE $g2 \leftarrow L[j]$
257 \STATE $B1[score] \leftarrow (g1,g2)$
263 \subsection{Features visualization}
265 The goal is to visualize results by building an evolutionary tree. All
266 core genes generated represent an important information in the tree,
267 because they provide ancestor information of two or more
268 genomes. Each node in the tree represents one chloroplast genome or
269 one predicted core and labelled as \textit{(Genes count:Family name\_Scientific
270 names\_Accession number)}. While an edge is labelled with the number of
271 lost genes from a leaf genome or an intermediate core gene. Such
272 numbers are very interesting because they give an information about
273 evolution: how many genes were lost between two species whether
274 they belong to the same lineage or not. To depict the links between
275 species clearly, we built a phylogenetic tree showing the
276 relationships based on the distances among genes sequences. Many tools
277 are available to obtain a such tree, for example:
278 PHYML\cite{guindon2005phyml},
279 RAxML{\cite{stamatakis2008raxml,stamatakis2005raxml}, BioNJ, and
280 TNT\cite{goloboff2008tnt}}. In this work, we chose to use
281 RAxML\cite{stamatakis2008raxml,stamatakis2005raxml} because it is
282 fast, accurate, and can build large trees when dealing with a large
283 number of genomic sequences.
285 The procedure used to built a phylogenetic tree is as follows:
287 \item For each gene in a core gene, extract its sequence and store it in the database.
288 \item Use multiple alignment tools such as (****to be write after see christophe****)
289 to align these sequences with each others.
290 \item Use an outer-group genome from cyanobacteria to calculate distances.
291 \item Submit the resulting aligned sequences to RAxML program to compute
292 the distances and finally draw the phylogenetic tree.
296 \centering \includegraphics[width=0.75\textwidth]{Whole_system} \caption{Overview
297 of the pipeline}\label{wholesystem}
300 \section{Implementation}
302 All the different algorithms have been implemented using Python on a personal computer running Ubuntu~12.04 with 6~GiB memory and
303 a quad-core Intel core~i5~processor with an operating frequency of
304 2.5~GHz. All the programs can be downloaded at \url{http://......} .
305 genes from large amount of chloroplast genomes.
309 \caption{Type of annotation, execution time, and core genes.}\label{Etime}
311 \begin{tabular}{p{2cm}p{0.5cm}p{0.25cm}p{0.5cm}p{0.25cm}p{0.5cm}p{0.25cm}p{0.5cm}p{0.25cm}p{0.5cm}p{0.2cm}}
313 Method & \multicolumn{2}{c}{Annotation} & \multicolumn{2}{c}{Features} & \multicolumn{2}{c}{Exec. time (min.)} & \multicolumn{2}{c}{Core genes} & \multicolumn{2}{c}{Bad genomes} \\
314 ~ & N & D & Name & Seq & N & D & N & D & N & D \\
316 Gene prediction & $\surd$ & - & - & $\surd$ & 1.7 & - & ? & - & 0 & -\\[0.5ex]
317 Gene Features & $\surd$ & $\surd$ & $\surd$ & - & 4.98 & 1.52 & 28 & 10 & 1 & 0\\[0.5ex]
318 Gene Quality & $\surd$ & $\surd$ & $\surd$ & $\surd$ & \multicolumn{2}{c}{$\simeq$3 days + 1.29} & \multicolumn{2}{c}{4} & \multicolumn{2}{c}{1}\\[1ex]
327 Table~\ref{Etime} presents for each method the annotation type,
328 execution time, and the number of core genes. We use the following
329 notations: \textbf{N} denotes NCBI, while \textbf{D} means DOGMA,
330 and \textbf{Seq} is for sequence. The first two {\it Annotation} columns
331 represent the algorithm used to annotate chloroplast genomes. The next two ones {\it
332 Features} columns mean the kind of gene feature used to extract core
333 genes: gene name, gene sequence, or both of them. It can be seen that
334 almost all methods need low {\it Execution time} expended in minutes to extract core genes
335 from the large set of chloroplast genomes. Only the gene quality method requires
336 several days of computation (about 3-4 days) for sequence comparisons. However,
337 once the quality genomes are well constructed, it only takes 1.29~minutes to
338 extract core gene. Thanks to this low execution times that gave us a privilege to use these
339 methods to extract core genes on a personal computer rather than main
340 frames or parallel computers. The lowest execution time: 1.52~minutes,
341 is obtained with the second method using Dogma annotations. The number
342 of {\it Core genes} represents the amount of genes in the last core
343 genome. The main goal is to find the maximum core genes that simulate
344 biological background of chloroplasts. With NCBI we have 28 genes for
345 96 genomes, instead of 10 genes for 97 genomes with
346 Dogma. Unfortunately, the biological distribution of genomes with NCBI
347 in core tree do not reflect good biological perspective, whereas with
348 DOGMA the distribution of genomes is biologically relevant. Some a few genomes maybe destroying core genes due to
349 low number of gene intersection. More precisely, \textit{NC\_012568.1 Micromonas pusilla} is the only genome who destroyes the core genome with NCBI
350 annotations for both gene features and gene quality methods.
352 The second important factor is the amount of memory nessecary in each
353 methodology. Table \ref{mem} shows the memory usage of each
354 method. In this table, the values are presented in megabyte
355 unit and \textit{gV} means genevision~file~format. We can notice that
356 the level of memory which is used is relatively low for all methods
357 and is available on any personal computer. The different values also
358 show that the gene features method based on Dogma annotations has the
359 more reasonable memory usage, except when extracting core
360 sequences. The third method gives the lowest values if we already have
361 the quality genomes, otherwise it will consume far more
362 memory. Moreover, the amount of memory, which is used by the third method also
363 depends on the size of each genome.
368 \caption{Memory usages in (MB) for each methodology}\label{mem}
371 \begin{tabular}{p{2.5cm}@{\hskip 0.1mm}p{1.5cm}@{\hskip 0.1mm}p{1cm}@{\hskip 0.1mm}p{1cm}@{\hskip 0.1mm}p{1cm}@{\hskip 0.1mm}p{1cm}@{\hskip 0.1mm}p{1cm}@{\hskip 0.1mm}p{1cm}}
373 Method& & Load Gen. & Conv. gV & Read gV & ICM & Core tree & Core Seq. \\
375 Gene prediction & NCBI & 108 & - & - & - & - & -\\
376 \multirow{2}{*}{Gene Features} & NCBI & 15.4 & 18.9 & 17.5 & 18 & 18 & 28.1\\
377 & DOGMA& 15.3 & 15.3 & 16.8 & 17.8 & 17.9 & 31.2\\
378 Gene Quality & ~ & 15.3 & $\le$3G & 16.1 & 17 & 17.1 & 24.4\\