=Paper=
{{Paper
|id=None
|storemode=property
|title=From Ontology for Genetic Interval (OGI) to Sequence Assembly -- Ontology applying to next generation sequencing
|pdfUrl=https://ceur-ws.org/Vol-559/Poster2.pdf
|volume=Vol-559
|dblpUrl=https://dblp.org/rec/conf/swat4ls/LinTS09
}}
==From Ontology for Genetic Interval (OGI) to Sequence Assembly -- Ontology applying to next generation sequencing==
https://ceur-ws.org/Vol-559/Poster2.pdf
From Ontology for Genetic Interval (OGI) to Sequence
Assembly
– Ontology Applying to Next Generation Sequencing
Yu Lin1, Hiroshi Tarui1, Peter Simons2
1. Genome Resource and Analysis Unit, Genomics Laboratory, Center for Development
Biology, RIKEN, Japan {linyu,tarui}@cdb.riken.jp
2. Department of Philosophy, Trinity College Dublin, Ireland psimons@tcd.ie
Abstract. We develop an OWL ontology: OGI (Ontology for Genetic Interval) for the
formalization of the genomic elements by defining them as a Genetic Interval. Based on
OGI’s definition of Genetic Interval Relations, which derived from the Allen interval
calculus, we attempt to represent the relationships among contigs and sequence data
from next generation sequencing. A real dataset generated from the bench has been
loaded and tested by using SPARQL for validating OGI. Although the dataset is small,
this semantic-based method provides a clue for assembly sequence. Evaluating this
method on a bigger dataset, both harmony and conflict of definitions with current
ontologies, such as SO (Sequence Ontology), need to be considered. OGI is available on
NCBO’s BioPortal website.
Keywords: OGI, ontology, genetic interval relations, next generation sequencing, 454
FLX sequencer, SuperContig
1 Background
The mother ontology for genetic interval is Ontology of Genetic Susceptibility
Factors (OGSF), which has been built as a modular ontology of the ontologies
for genetic susceptibility to disease [1]. OGSF includes three ontologies:
Ontology of Genetic Susceptibility Factors (OGSF), Ontology of Glucose
Metabolism Disorder (OGMD) and Ontology of Geographical Region (OGR).
When we discovered that the co-localization of genetic susceptibility factors
(such as SNPs) with the gene is the major criterion for determining a
susceptibility gene, we developed another ontology: Ontology for Genetic
Interval (OGI, http://bioportal.bioontology.org/ontologies/40117). The purpose of
OGI is to formalize the genomic elements, including genes, mutations mRNAs,
and all kinds of genomic structures which are essential to current genomics
research.
2 Introduction of OGI
In current molecular biology research, the gene and other genomic elements
have been modeled as a sequence, which is represented as the combination of A,
T, C and G in a linear form. Ontology for Genetic Interval described the Genetic
Interval as a subclass of Biological Interval, which is a “spatial continuous
physical entity which contains ordered biological sets (DNA segment, Nucleic
Acid Base Residue, RNA segment, Protein segment) between two boundaries:
start boundary and end boundary on a chromosome, RNA or protein”.[2] The
difference between Genetic Interval and Genetic Sequence is that Genetic
Interval holds not only the primary linear sequence information but also the 3-
D structure information of a given Genetic Interval. However, since the current
capability of genomics research is limited to the linear information of a Genetic
Interval, the genetic sequence is taken into account as a simplified model of
genetic interval.
In OGI, we define a sequence as a specific kind of group (collective) within
the methodological and ontological constraints of nominalism. In order to
conceptualize the notion of sequence we start from logic, which is an
indispensable part of constructing an ontology. Then we define the Genetic
Interval Relations by borrowing ideas from the Allen interval calculus (Table1).
Table 1. Genetic Interval Relations
Relations in Allen Illustration Relations of Genetic Interval
Interval
XX isLocatedAfter (yLAx)
XmY isAdjacentBefore (xABy)
YmiX isAdjacentAfter (yAAx)
XoY isOverlapStartWith (xOSy)
YoiX isOverlapEndWith (yOEx)
XsY isStartsWith (xSWy,ySWx)
YsiX (symmetric property)
XdY isContainedIn (xCIy)
YdiX (transitive property)
XfY isEndWith (xEWy, yEWx)
YfiX (symmetric property)
X=Y isEqualTo (xEy, yEx)
(symmetric property)
!" isReverseCompleteOf (xRCy)
#" (symmetric property)
3 Next Generation Sequencing and Sequence Assembly
According to wikipedia, “The term DNA sequencing refers to sequencing
methods for determining the order of the nucleotide bases—adenine, guanine,
cytosine, and thymine—in a molecule of DNA.” [3] State-of-the-art next-
generation sequencing platforms such as the Roche-454 GS FLX, Illumina
Genome Analyzer and ABI SOLiD provide high-throughput and high-speed
technology to read the nucleotide bases of samples. However, the reading length
generated by such sequencers is very short: ~400bp by Roche-454 FLX titanium,
~75bp by Illumina, and ~50bp by ABI SOLiD. The previous and widely used
sequencer is Sanger 3730 series, by which a reading length up to 2000bp can be
obtained.
Usually, a sequencing project is taken by randomly cutting the target
sequence into smaller fragments, which is the so-called “shotgun method”; and
then after next generation sequencers obtain the readings, a computer will
attempt to put all readings together to give the whole view of the target
sequence. Thus, the assembly procedure for combining the readings either by
mapping or de novo method is essential. How to get a closest overall picture of
the target sequence is a key issue for both bioinformaticians and
biotechnologists.
4 A Practice of Using OGI to Assemble the Readings
4.1 Method
In this experiment, the length of target sequence is up to 60k bp. The shotgun
sequencing method was applied to get the fragments of sequences; after
amplification by PCR, all the fragments were then mixedshothe up for building
a DNA library for running on a 454 FLX sequencer. After we got the raw
readings dataset from the 454 FLX sequencer, we used the software package
provided by 454 Life Sciences Corporation. Since the sample comes from a
model organism, we chose the GS Reference Mapper Application from the
package to assemble reading. All the ~300bp readings were assembled as 5
contigs. (Contig: a group of overlapping readings derived from a single genetic
source.) Comparing the contigs with the reference genomic sequence by using a
similarity alignment method, we found out that there are four gaps between
those contigs. According to the length of the Gaps and Contigs, a resequencing
region flanking the ending of Contig2, Gap2, Contig3, Gap3, Contig4, Gap4 and
the beginning of Contig5 has been decided. (Fig. 1. )
Fig. 1. Contigs and Gaps
Except for Gap1, which can be filled by manually checking the sequence, we
resequenced the resequencing region above by the shotgun method using a
Sanger 3730 sequencer. After repeating the alignment and reference sequence
comparison steps as in the previous stage, relationships between those Contigs
and Gaps were represented using Genetic Interval Relations from OGI. Then,
all these entities (Contig and Gap) and their relations were manually populated
into OGI under the software environment of Protégé 4.0.
4.2 Result
4.2.1 Representing the relations between Contigs and Gaps
!"#$%&'(%)*"+,$-./-0"1-(!"#$%&2
!"#$%&2(%)*"+,$-./-0"1-(!"#$%&3
!"#$%&3(%)*"+,$-./-01-(!"#$%&4
!"#$%&4(%)*"+,$-./-01-(!"#$%&5
6,72(%)8.9,+-#$/-0"1-(!"#$%&3
6,73(%)8.9,+-#$/-0"1-(!"#$%&4
6,74(%)8.9,+-#$/-0"1-(!"#$%&5
!"#$%&'(%):;-1<,7=$,1$>%$?(!"#$%&2
!"#$%&2(%):;-1<,7=,1$>%$?(@-;@-)-AB'2
!"#$%&3(%)!"#$,%#-.C#(@-;@-)-AB'2
6,72(%)!"#$,%#-.C#(@-;@-)-AB'2
@-;@-)-AB'2(%):;-1<,7D#.>%$?(6,73
!"#$%&4(%):;-1<,7=,1$>%$?(@-)-AD'2
6,74(%)!"#$,%#-.C#(@-)-AD'2
@-)-AD'2(%):;-1<,7D#.>%$?(!"#$%&5
!"#$%&3(%):;-1<,7=$,1$>%$?(@-;@-)-AE'2
6,73(%)!"#$,%#-.C#(@-;@-)-AE'2
!"#$%&4(%)!"#$,%#-.C#(@-;@-)-AE'2
6,74(%)!"#$,%#-.C#(@-;@-)-AE'2
@-;@-)-AE'2(%):;-1<,7=$,1$>%$?(!"#$%&5
@-;@-)-AE'2(%)@-;-1)-!"F7<-$-:0(@-)-AE'2
@-;@-)-AB'2(%)@-;-1)-!"F7<-$-:0(@-)-AB'2
4.2.2 Axioms which are built in Protégé 4.0 as follows:
Axiom1: if Z isOverlapStartWith X and X isContainedIn Y,
then,
Z isOverlapEndWith Y => Y isOverlapStartWith Z
Axiom2: if Y isOverlapEndWith X and X isContainedIn Z, then,
Y isOverlapStartWith Z => Z isOverlapEndWith Y
Using the pellet reasoner from Protégé 4.0, the following relations are inferred
by the software:
@-;@-)-AB'2(%):;-1<,7=$,1$>%$?(@-;@-)-AE'2
@-;@-)-AE'2(%):;-1<,7=$,1$>%$?(@-)-AD'2
4.2.3 SuperContig analysis
A SuperContig means complete nucleotide sequence information of the sample,
in our case, the 60kbp’s target sequence. The SuperContig should start with
Contig1 and end with Contig5, which also means that we are looking for a path
which starts from Contig1 and ends with Contig5.
=G7-1!"#$%&(
(((((((((%)=$,1$>%$?(!"#$%&'
(((%)D#.>%$?(!"#$%&5
SuperContig is an accumulation of all the assembled Contigs, and all the
subsets relations in this SuperContig must be the same as each other:
isOverlapStartWith .
Fig. 2. Path for building a SuperContig
(The red dashed line shows the path to generate a SuperContig in Fig.2).
• Using an open source reasoning tool kit LSW[4], which uses Pellet as the
underlying reasoner, we ran SPARQL codes and got following results:
SELECT ?x ?y ?z
WHERE {
?x OGI:isOverlapStartWith OGI:Contig5 .
?y OGI:isOverlapStartWith ?x .
?z OGI:isOverlapStartWith ?y .
OGI:Contig1 OGI:isOverlapStartWith ?z . }
Results: OGI:RevReseqF12 OGI:RevReseqD12 OGI:Contig2
• Thus, the SuperContig were constructed by (in the correct order):
!"#$%&'(%):;-1<,7=$,1$>%$?(!"#$%&2
!"#$%&2(%):;-1<,7=,1$>%$?(@-;@-)-AB'2
@-;@-)-AB'2(%):;-1<,7=$,1$>%$?(@-;@-)-AE'2
@-;@-)-AE'2(%):;-1<,7=$,1$>%$?(!"#$%&5
! SuperContig will be generated by overlapping the nucleotide sequence of
Contig1, Contig2, RevReseqD12, RevReseqF12, and Contig5.
5 Conclusion
In this study, we have populated a small dataset of contigs assembled from the
454 FLX sequencing readings to OGI; the purpose is to construct a SuperContig
which can give the complete nucleotide sequence information of the target
sequence. Pellet reasoner, logic rules and SPARQL language were applied for
finding the path for building SuperContig. Being able to form one SuperContig
out of all the readings is an essential step in sequencing method. Many
assembly tools are using either mathematical or object-oriented methods to
construct the SuperContig. Here we tried to follow the semantic method, which
helps people understand the relations of contigs, especially those resequencing
contigs or fragments generated by different sequencers, in our case, both next
and first generation sequencer used for a deep and accurate sequencing.
Since our method applied to the limited contigs generated by standard software
rather than the huge readings, it is practical and reasonable. However, when
dealing with a whole genome, especially those which have no reference genome
sequences to compare with, more contigs will complicate the assembly. A better
solution will be needed for accelerating the capability of reasoning and
performance.
Another issue is that the overlap between OGI and current Sequence
Ontology (SO) [5] is unavoidable. Although not yet discussed with the
consortium of Sequence Ontology, OGI contributes to SO by providing the
relations described in this paper. Authors of OGI are planning to merge SO into
OGI by adopting the terms from SO, and it will be important to report a
detailed mapping between those two ontologies.
References
1. Lin Y., Sakamoto N.: Ontology of Genetic Susceptibility Factors to Diabetes Mellitus
(OGSF-DM). Interdisciplinary Ontology Proceedings of the First Interdisciplinary
Ontology Meeting. 99 --104 (2008)
2. Lin Y., Sakamoto N.: Genome,Gene,Interval and Ontology. Interdisciplinary Ontology
Proceedings of the Second Interdisciplinary Ontology Meeting. 25—34 (2009)
3. http://en.wikipedia.org/wiki/DNA_sequencing
4. http://esw.w3.org/topic/LSW LSW is an open source set of lisp tools for working with
OWL and SPARQL using the Pellet reasoner. It was initially written by Alan
Ruttenberg, but is starting to accumulate contributions from others.
5. http://www.sequenceontology.org/ The Sequence Ontology is a set of terms and
relationships used to describe the features and attributes of biological sequence. SO
includes different kinds of features which can be located on the sequence.