Fact and Task Oriented System for Genome
Assembly and Annotation
Luciano A. Digiampietri, Julia M. Perdigueiro, Aloisio J. de Almeida Junior,
Daniel M. Faria, Eric H. Ostroski, Gustavo G.L. Costa, and Marcelo C. Perez
Instituto de Computa¸ c˜ ao, Universidade Estadual de Campinas,
CP 6176, Campinas, SP 13084-971 BRAZIL
Abstract. We present a preliminary description and results of a system
to help the curation of genome assembly and annotation. Standard tools
are used for these tasks, and our methodology focuses on user guidance,
data visualization and integration, and data browsing aspects.
The usual concern of most of activities, tools and infrastructure related to ge-
nomic analyses is with computer systems functionality. Many systems are devel-
oped in an ad hoc way following only functional requirements. This development
methodology pays little attention to characteristics like user interface and usabil-
ity. We have developed a simple methodology to make the user-interaction part
of genome assembly and annotation more user-friendly and therefore more effec-
tive. Based on this methodology we have implemented a web-based prototype.
This prototype is being used as the main tool for the assembly and annotation of
the Xanthomonas axonopodis pv aurantifolii strains B and C genomes at LBI 
with the support of USP  and UNESP .
2 System Development Methodology
The system presented here was developed following a generic methodology spec-
ified by us at LBI. This methodology allows the development of any compu-
tational infrastructure which requires a flow of activities and that provides
data mining and visualization mechanisms. This methodology has the following
phases: (i) identification and description of tasks to be done; (ii) description of
facts to be considered; (iii) development of fact analysis and visualization tools;
(iv) development of examples or tutorials on how to execute each task; (v) devel-
opment of tools for accomplishing the tasks. We have applied this methodology
to improve a genome assembly and annotation system used at our laboratory.
Facts are characteristics observed in the set of available data. Facts are the
basis for all the analysis and conclusions which will be made during assembly
and annotation. Tasks are actions which must be executed (automatically or
J.C. Setubal and S. Verjovski-Almeida (Eds.): BSB 2005, LNBI 3594, pp. 238–241, 2005.
c ?Springer-Verlag Berlin Heidelberg 2005
Fact and Task Oriented System for Genome Assembly and Annotation 239
Table 1. Assembly tasks and facts
Contigs projection on the
Supercontigs management contigs, links, gap closures and alignments to the refer-
to besubcloned and
set of reads, phrap and genscaff results
reads from the same insert found on different contigs
alignment between the reference and the target genomes
reads and links information
manually) with the objective of getting closer to the desired solution. For ex-
ample, a set of facts can be observed in the result of the phrap  assembly
and postprocessing by the genscaff program , such as a possible link between
contig x and contig y. A task must obtain conclusions about the facts, for in-
stance, to conclude whether contigs x and y are adjacent or not. For each kind
of fact, data analyses and visualization tools were developed to ease the under-
standing and the making of a decision. Some examples of genome assembly tasks
are: contigs management, links management, selection of clones to be subcloned
and sequenced, comparison between the target and the reference genomes and
supercontig management (supercontig is a set of linked contigs).
Figure 1 shows some of the graphical results of our tools (showing contig,
supercontig, link and projection with reference genome information). All figures
are automatically generated and have hyperlinks to allow easy data browsing.
One of the most complex tasks during genome assembly is to decide whether
two contigs are linked or not. Our system used the following facts to help in
decision making: (1) links between those contigs; (2) conservation of the order
regarding the reference genome (based on alignment against a reference genome);
Fig.1. Supercontig information: contigs, links, gaps and projection over the reference
240 L.A. Digiampietri et al.
and (3) bionformatics gap closure (a sub-assembly using only reads in a particu-
lar region that successfully closes a gap). By integrating these facts, our system
facilitates the curatorial part of the genome assembly process, decreasing the
need for new sequencing.
3 Results, Conclusions and Future Work
Complex information systems that require intense user interaction deserve spe-
cial care on user-related issues, such as usability and interface. Large-volume
data processes, such as genome assembly and annotation, require special care
on data presentation, through graphic visualizations, data summaries and data
integration. We have briefly described a simple methodology that helped us cre-
ate a web-based system that allowed us to achieve good results in a genome
assembly process. The detailed description of each one of the tasks and facts, as
well as the specification of tutorials or examples for each task, makes possible a
more conscious, easy and systematic use of the system.
The system proposed is being used on the Xanthomonas axonopodis pv au-
rantifolii strains B and C genomes assembly and annotation. Before the work
described in this paper, these two genomes were being assembled using a tradi-
tional system, which had no specific computational help for assembly curators.
The use of our system showed quantitative and qualitative gains with respect
to previous assembly results. The main gains were: (i) all data is integrated in
a database management system (DBMS), making it possible to make efficient
queries to every object involved in the project; (ii) low training cost of new
assembly and annotation team, due to tutorials developed for the execution of
each task; (iii) greater assembly efficiency through a better use of data. The
most important practical conclusion of this case study was the reduction on the
number of supercontigs without the need of new sequencing, causing greater
genome coverage. Table 2 compares the results obtained by our system to the
ones available before we put our system to use. This table shows that thorough
our system we obtained better results on every analyzed characteristic, refining
the assembly and being more efficient on the use of available data.
As a future step we intend to package tools (making them more generic and
reusable), and extending the system for dealing also with comparative genomics.
Table 2. Comparison between previous results and results from our approach
Our results Situation
Number of supercontigs
Total number of contigs in the supercontigs
Average number of contigs by supercontig
Number of base pairs on supercontigs
Valid links on supercontigs
Number of new closed gaps
Fact and Task Oriented System for Genome Assembly and Annotation 241 Download full-text
Another future work is the development and usage of ontology to publish data
on the Web through XML , increasing interoperability.
More detailed descriptions and tools can be obtained through e-mail contact
with LBI: firstname.lastname@example.org.
Acknowledgements. The work described in this paper was partially financed
by CAPES, Fundecitrus and MCT-PRONEX SAI project. The Laboratory of
Bioinformatics, in which most of this work was done, is partly funded by FAPESP.
We acknowledge the others researchers that had worked in the sequencing and
assembling of the genomes presented here.
1. Departamento de Bioqu´ ımica, Instituto de Qu´ ımica, University of Sao Paulo.
2. Extensible Markup Language (XML) 1.0 (Third Edition) (2004).
3. Green P. Phrap: phragment assembly program. http://www.phrap.org.
4. Laboratory for Bioinformatics (LBI), Institute of Computing, University of Camp-
5. Laborat´ orio de Bioqu´ ımica e Biologia Molecular (LBM), UNESP.
6. Setubal, J. and Werneck, R.: A program for building contig scaffolds in double-
barreled shotgun genome sequencing. Technical Report IC-01-05, Institute of Com-
puting, Unicamp, 2001.