Skip to content
Snippets Groups Projects
Commit 30ba85ef authored by Laros's avatar Laros
Browse files

Updated the Mutalyzer 2.0 paper. 

git-svn-id: https://humgenprojects.lumc.nl/svn/mutalyzer/trunk@645 eb6bd6ab-9ccd-42b9-aceb-e2899b4a52f1
parent 823f5bc0
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
doc/Mutalyzer 2.0/paper.bbl
+ 6
− 0
View file @ 30ba85ef
......@@ -14,6 +14,12 @@ J.F.J. Laros, A.~Blavier, J.T. den Dunnen, and P.E.M. Taschner.
{National Center for Biotechnology Information}.
\newblock \begin{small}\texttt{http://www.ncbi.nlm.nih.gov/}\end{small}.
\bibitem{DBSNP}
S.T. Sherry, M.H. Ward, M.~Kholodov, J.~Baker, L.~Phan, E.M. Smigielski, and
K.~Sirotkin.
\newblock {dbSNP}: the {NCBI} database of genetic variation.
\newblock {\em Nucleic Acids Research}, 29(1):308--11, January 2001.
\bibitem{Mutalyzer}
M.~Wildeman, E.~van Ophuizen, J.~T. den Dunnen, and P.~E. Taschner.
\newblock Improving sequence variant descriptions in mutation databases and
......
doc/Mutalyzer 2.0/paper.tex
+ 212
− 106
View file @ 30ba85ef
......@@ -37,7 +37,7 @@ one.
%Distributed setup, multiple servers can be used, cache and database are
%synchronised.
The suite is logically divided in modules which can be combined to make several
The suite is logically divided in modules which, combined make up several
interfaces.
\begin{table}[]
......@@ -46,69 +46,203 @@ interfaces.
\begin{tabular}{l|l}
name & function\\
\hline
HGVS parser & Check the HGVS nomenclature.\\
HGVS parser & Check the HGVS syntax.\\
Retriever & Retrieve a reference sequence.\\
Crossmapper & Convert positions from \texttt{g.} to \texttt{n.} or
Cross\-mapper & Convert positions from \texttt{g.} to \texttt{n.} or
\texttt{c.} and vice versa.\\
Database & Interface to Mutalyzer's internal database.\\
Mutator & Apply variants to a reference sequence.\\
GenRecord & Generalisation of reference sequences.\\
VariantChecker & Semantic checks.
\end{tabular}
\end{center}
\label{tab:modules}
\end{table}
In Table~\ref{tab:modules} we see a list of core modules in the Mutalyzer
suite, the functionality of each module is described in the following sections.
A list of core modules of the Mutalyzer suite is shown in
Table~\ref{tab:modules}, the functionality of each module is described in the
following sections.
Database:
\begin{itemize}
\item Mapping of transcripts and genes.
\item Cache administration.
\item Batch checker.
\end{itemize}
\subsubsection{HGVS parser} \label{subsubsec:parser}
The formalisation of the HGVS nomenclature~\cite{hgvs_bnf}, made it possible to
implement a context free grammar parser that encompasses the complete
nomenclature, including nesting and other newly added features. The
\emph{parser} module is the implementation of this context free parser. The
major functionalities of this module are recognising all syntactically valid
descriptions and generating a \emph{parse tree}, which is used to interpret the
variant description.
\subsubsection{Retriever} \label{subsubsec:retriever}
To verify the validity of variant descriptions, the reference sequence file on
which the variant is described is needed. Reference sequence files contain the
reference sequence and annotation about transcripts, CDSs, exons, etc. These
reference sequences are retrieved from either the NCBI (for GenBank files) or
the EBI (in case of LRG files). The \emph{retriever} module is responsible for
the communication with these repositories, as well as local administration of
downloaded files.
For performance reasons, reference sequences are stored in a cache that resides
on the server that Mutalyzer runs on. Administration of the cache is done by
the \emph{database module} described in Section~\ref{subsubsec:db}.
\subsubsection{Cross\-mapper} \label{subsubsec:crossmap}
The HGVS nomenclature has different \emph{positioning systems} (\texttt{g.},
\texttt{c.}, etc.), the non-trivial conversion between these systems is done by
the \emph{cross\-mapper} module. By facilitating the bidirectional conversion
from \texttt{c.} and \texttt{n.} to \texttt{g.} and \texttt{m.}, this module
can be used to convert positions between transcripts.
% UD_135074263017:g.8000del
% NC_000011.9:g.111949587del
% UD_135074263017(PIH1D2_v001):c.-4915del
% UD_135074263017(C11orf57_v001):c.170+548del
% UD_135074263017(TIMM8B_v001):c.*6432del
% UD_135074263017(TIMM8B_v002):n.*5937del
% UD_135074263017(SDHD_v001):c.-8045del
\subsubsection{Database} \label{subsubsec:db}
Since much of the retrieved information is reusable and quite some
administration of different .. needs to be done, Mutalyzer uses a database to
store \ldots
To facilitate the conversion of chromosomal coordinates to gene-oriented ones,
a number of \emph{mapping databases} were created. These databases contain the
locations of all exons of all transcripts, together with the coordinates of the
CDS. With this information the Mutalyzer Crossmapper module (described in
Section~\ref{subsubsec:crossmap} can convert all descriptions on transcripts
(\texttt{NM}) to chromosomal ones and vice versa.
For various reasons, discussed in Section~\ref{subsubsec:genrecord}, the link
between a transcript and a protein accession number needs to be known, e.g.,
the product of \texttt{NM\_000193.2} is \texttt{NP\_000184.1}. This link is
retrieved from the NCBI and for performance reasons they are stored in the
database.
When placing a reference sequence in the cache, the md5sum of the file is
calculated and stored in the database, together with information about its
source. If the reference sequence is a slice of a chromosome or contig, the
parameters used to make this slice are stored. For other methods of uploading
all information necessary to retrieve the reference sequences once it is
removed from the cache is stored. This way of cache administration also
prevents re-uploading of reference sequences that are already known to
Mutalyzer, so in principle no reference sequence will be stored under different
accession numbers.
When clients submit a batch job to the \emph{batch queueing system}, described
in Section~\ref{subsec:batch}, all necessary information about the job, as well
als the content of the job itself, is stored in the database. Every client is
assigned to their private queue and tasks are selected from these queues in a
round-robin way. This approach guarantees that a small job will finish in a
short amount of time, regardless of the size of other scheduled jobs (only
depending on the amount of scheduled jobs).
\subsubsection{Mutator}
In order to perform any contextual checks, the retrieved reference sequence
must be modified to assess the validity of a variant description. To facilitate
the correction of ill formed but interpretable descriptions, a simulation of
the variant is made by the \emph{mutator} module. By tracking the position
shifts resulting from the simulation of partial variants an allele description
can be simulated.
\subsubsection{GenRecord} \label{subsubsec:genrecord}
In order to be able to work with different types of reference sequences
(GenBank, LRG, EMBL, \ldots), an abstraction of reference sequence files is
needed. The \emph{Genrecord} module is an implementation of such an
abstraction.
Since the annotation if the various reference sequence files are not strict,
multiple annotation enrichment schemes are needed to standardise the
information contained in them. The reference sequence annotation enrichment
consists mainly of the linking of annotated transcripts a \emph{coding
sequence} (CDS) and protein. In many cases, (especially in GenBank files) there
is no direct link between a transcript and its CDS, making it impossible to
reconstruct the layout of the transcript and thereby the biological effect of a
variant. Therefore an extensive set of methods is developed to find this link.
For each transcript we find all CDSs that are consistent with the transcript in
question, this can be considered as quality control of the annotation. In
general this does not uniquely assign a CDS to a transcript but it does narrow
down the number of possibilities. A link between the annotated mRNA and protein
accession numbers as described in Section~\ref{subsubsec:db} is a reliable way
of resolving ambiguities, so usage of this information is preferred when
available. An other reliable, but frequently missing, way of linking mRNA to a
CDS is by using the \emph{locus} tag.
As a last resort, the product tags of both mRNA and CDS may be used, for the
set of CDSs, we determine the longest common prefix and suffix to find the
consecutive words that are not shared between all product tags. We do the same
for the set of transcripts. If these operations lead to a list that uniquely
identifies the transcripts and CDSs, we proceed to match the two lists.
\begin{table}[]
\caption{Usage of the product tag.}
\begin{center}
\begin{tabular}{lll}
Type & Accession number & Product\\
\hline
CDS & \texttt{NM\_000109.3} & dystrophin, transcript variant Dp427c
protein\\
CDS & \texttt{NM\_004006.2} & dystrophin, transcript variant Dp427m
protein\\
CDS & \texttt{NM\_004009.3} & dystrophin, transcript variant Dp427p1
protein\\
mRNA & \texttt{NP\_000100.2} & dystrophin Dp427c isoform protein\\
mRNA & \texttt{NP\_003997.1} & dystrophin Dp427m isoform protein\\
mRNA & \texttt{NP\_004000.1} & dystrophin Dp427p1 isoform protein\\
\end{tabular}
\end{center}
\label{tab:product}
\end{table}
Consider the example in Table~\ref{tab:product}, the identifying words in the
set of CDSs are Dp427c, Dp427m and Dp427p1. The same words are found in the set
of transcripts so this information can be used to resolve ambiguity.
The methods described in this section are tried in order and the method that
finally resolved the ambiguity is reported by Mutalyzer.
\subsection{Name Checker} \label{subsec:namecheck}
The formalisation of the HGVS nomenclature~\cite{hgvs_bnf}, made it possible to
implement a context free grammar parser that encompasses the complete
nomenclature, including nesting and other newly added features. Although the
semantic checks have not been implemented for the complete nomenclature, we
have implemented the recognition of allele descriptions. These descriptions
are, apart from simple variants consisting of one change, the most occurring
ones. With allele descriptions we can describe a large number of changes, i.e.,
all descriptions that need no information from outside the reference sequence.
% Parse description
Although the contextual checks have not been implemented for the complete
nomenclature, the recognition of allele descriptions has been implemented.
These descriptions are, apart from simple variants consisting of one change,
the most occurring ones. With allele descriptions one can describe a large
number of changes, i.e., all descriptions that need no information from outside
the reference sequence.
We define a \emph{raw variant} as an elementary variant, e.g., a substitution,
deletion, insertion, etc. An \emph{allele description} is obtained by
concatenation of raw variants. Consider the following description
\texttt{NM\_002001.2:c.[10del;22C>T;101\_119inv]}. The part between brackets is
the allele description, which consists of raw variants separated by a semicolon
(\texttt{10del}, \texttt{22C>T} and \texttt{101\_119inv}).
After the description is parsed, the \emph{reference sequence}, which is part
of the description, is retrieved from a reference sequence repository (e.g.,
the NCBI~\cite{NCBI} or the EBI). The reference sequence (in either GenBank or
LRG format) is then parsed and any missing data is retrieved to prepare for the
next step. See Section~\ref{sec:enrichment} for a complete description.
With the parsed variant description and the information from the reference sequence, the variant can be simulated to produce the observed
sequence. In this simulation all raw variants are visualised and checked. The
checks on the raw variants is extensive.
First, we check whether the minimal description is used. In case of a
\begin{center}
\texttt{NM\_002001.2:c.[10del;22C>T;101\_119inv]}
\end{center}
The part between brackets is
the allele description, which consists of raw variants (\texttt{10del},
\texttt{22C>T} and \texttt{101\_119inv}) separated by a semicolon.
After the description is parsed, the \emph{reference sequence}, uniquely
identified by the accession number, which is part of the description, is
retrieved from a reference sequence repository (e.g., the NCBI~\cite{NCBI} or
the EBI). The reference sequence (in either GenBank or LRG format) is then
parsed and any missing data (described in Section~\ref{subsubsec:genrecord}) is
retrieved to prepare for the next step.
With the parsed variant description and the information from the reference
sequence, the variation can be simulated to obtain the observed sequence. In
this simulation all raw variants are visualised and checked. The checks on the
raw variants is extensive.
First, Mutalyzer checks whether the minimal description is used. In case of a
\texttt{delins}, one can for example add reference bases to the inserted
sequence, adding nothing to the description. If, for example, the reference
sequence is \texttt{AACGTAA}, we can define the following deletion-insertion:
\texttt{3\_4delinsTT}, resulting in an observed sequence of \texttt{AATTTAA}.
The same result will be obtained if we define the variant as
sequence, adding no information to the description. If, for example, the
reference sequence is \texttt{AACGTAA}, we can define the following
deletion-insertion: \texttt{3\_4delinsTT}, resulting in an observed sequence of
\texttt{AATTTAA}. The same result will be obtained if we define the variant as
\texttt{2\_6delinsATTTA}. This latter description can be minimised by
calculating and removing the \emph{longest common prefix} and the \emph{longest
common suffix} of the deleted and the inserted sequence.
In case of an inversion, a prefix of the inversion can be the reverse
complement of a suffix, we call this a \emph{partial palindrome}. The
complement of its suffix, i.e., a \emph{partial palindrome}. The
description of an inversion is minimised in a similar way as described above.
\begin{table}
......@@ -128,7 +262,7 @@ description of an inversion is minimised in a similar way as described above.
After the minimisation step, a disambiguation scheme is used to check whether
the type of the raw variant is correct. In Table~\ref{tab:typedisambiguation}
we see which variant types can possibly be written as a simpler type.
it is shown which variant types can possibly be written as a simpler type.
Finally, a deletion or insertion is shifted to the most 3' position possible.
We use a \emph{rolling} algorithm that takes circular permutations of the
......@@ -136,64 +270,63 @@ deleted or inserted sequence into account. If for example, we insert the
sequence \texttt{TCCA} in a reference sequence \texttt{CATC}, the
algorithm will correct the description \texttt{2\_3insTCCA} to
\texttt{3\_4insCCAT}. Note that this method works for both the forward as well
as the reverse strand. If a gene resides on the reverse strand, the position
as the reverse strand, so if a gene resides on the reverse strand, the position
will be shifted in the opposite direction of that of the genomic one.
Furthermore, if an insertion or a deletion is described on a transcript, the
position will not be shifted over a splice site.
%Optional arguments (\texttt{10\_12delAAT}) are checked.
After the simulation of the variant, we have the observed sequence. We use this
observed sequence to do basic effect prediction.
After the simulation of the variation, we have the observed sequence. We use
this observed sequence to do basic effect prediction.
For all the annotated transcripts in the reference sequence, the corrected
variant description is shown, as well as a list of protein descriptions. Each
of the DNA variant descriptions can be selected for a more detailed analysis.
In the detailed analysis the reference protein and the variant protein is
visualised and the area of change is highlighted. For the selected transcript,
a list of exons start and end positions are given, as well as the CDS start and
a list of exon start and end positions are given, as well as the CDS start and
end positions.
For all raw variants, effects on restriction sites are calculated. A table is
generated that contains the number of the raw variant, a list of removed
restriction sites and a list of added restriction sites.
\texttt{Martijn}
Deletion of exons as well as partial exons (resulting in a fusion exon) is
supported.
Gives informative warnings when a variant is near a splice site.
%Warns when a change has no effect (\texttt{10A>A}, inversion of a palindrome,
%etc.).
Supports ``fuzzy'' positions.
\subsection{Syntax Checker}
The \emph{Syntax Checker} is an interface to the HGVS parser, described in
Section~\ref{subsec:namecheck}, only. This interface will only return whether
or not the syntax of a variant description is correct, no semantic check is
performed. If no reference sequence is available, one might be restricted to
checking the syntax only. An other reason for using the syntax checker is to
check large quantities of descriptions in a small amount of time. Since there
is no communication with reference sequence repositories, this check is
extremely quick.
If no reference sequence is available, or if large quantities of
descriptions in a small amount of time need to be checked, it might be
desirable to only check the syntax. The \emph{Syntax Checker} is an interface
to the HGVS parser, which will return whether or not the syntax of a variant
description is correct. No contextual check is performed. Since there is no
communication with reference sequence repositories, this check is extremely
quick.
\subsection{Position Converter}
The \emph{position converter} is an interface to the HGVS parser, the
crossmapper and the database. With this interface, we can convert a description
that uses a RefSeq transcript as reference sequence to a description on a
chromosomal reference sequence and vice versa. The mapping information is
retrieved daily from the NCBI. Currently Human genome build 18 and 19 are
supported.
cross\-mapper and the database. With this interface, we can convert a
description that uses a RefSeq transcript as reference sequence to a
description on a chromosomal reference sequence and vice versa. The mapping
information is retrieved daily from the NCBI. Currently Human genome build 18
and 19 are supported.
This interface can be used to quickly convert variants found by a high
throughput screening, e.g., an NGS experiment. This enables people to annotate
their NGS experiments with informative HGVS descriptions. An other use of this
interface is to convert (or lift over) a description from one transcript to an
interface is to convert (lift over) a description from one transcript to an
other, or to transcripts of other (overlapping) genes. Finally, by using
transcripts that are mapped to both hg18 as well as to hg19, we can convert a
chromosomal description from one build to an other. Potentially, we can even
lift over descriptions to other species.
chromosomal description from one build to an other. Potentially, descriptions
can be lifted over to other species, provided cross-species transcript
annotation is available.
\subsection{SNP Converter}
For converting a DbSNP~\cite{DBSNP} id to an HGVS description, the \emph{SNP
......@@ -201,6 +334,8 @@ converter} can be used. This interface retrieves the annotated HGVS
descriptions from the NCBI.
\subsection{Name Generator}
\texttt{Gerben}
Educational interface for those who are not familiar with the HGVS
nomenclature.
......@@ -208,35 +343,18 @@ Constructed variant description can be checked (clickable) with the name
checker.
\subsection{Reference File Loader}
For reference sequences unknown to the NCBI or EBI, we have created the
\emph{reference file loader}.
\begin{enumerate}
\item Upload a local file. \label{item:local}
\item Download a reference sequence by supplying a URL.
\item Retrieve part of the reference genome for a (HGNC) gene symbol
\begin{itemize}
\item Most recent build is used for the organism.
\item The orientation of the slice is selected automatically.
\item Flanking ranges.
\end{itemize}
\item Retrieve a range of a chromosome by accession number
\begin{itemize}
\item Choose orientation.
\end{itemize}
\item Retrieve a range of a chromosome by name
\end{enumerate}
Reference sequences are stored in a cache.
MD5sum is used to identify the file.
\begin{itemize}
\item Prevents re-uploading, the same UD is returned.
\item Enables the retrieval of reference sequences after it has vanished from
the cache, except for~\ref{item:local}.
\end{itemize}
\subsection{Batch Jobs}
To support reference sequences unknown to the NCBI or EBI, we implemented a
\emph{reference file loader}. This interface can be used to upload a local
file, upload a file by supplying an URL, or to slice a chromosome or contig.
A slice can be made directly by supplying the name or accession number of a
chromosome or contig, the location of the slice and the orientation.
Additionally, a gene name in combination with the name of an organism can be
used to select the slice automatically. In this mode, the most recent build of
the organism in question will be used, the orientation is selected
automatically, the size of the flanking regions (5' and 3') can be modified.
\subsection{Batch Jobs} \label{subsec:batch}
For the Name Checker, Syntax Checker, Position Converter and SNP Converter.
Formats (automatically detected):
......@@ -313,18 +431,6 @@ Nesting.
\section{Webservices}
\LTXtable{\textwidth}{webservices.tex}
\section{Annotation enrichment} \label{sec:enrichment}
The reference sequence annotation enrichment consists mainly of the linking of
annotated transcripts to their \emph{coding sequence} (CDS) and protein. In
many cases, (especially in GenBank files) there is no direct link between a
transcript and its CDS, making it impossible to reconstruct the layout of the
transcript and thereby the biological effect of a variant. Therefore we have
developed an extensive set of methods to accomplish this link. First we select
the CDSs that are consistent with a certain transcript, then we try to find a
connection between the CDS and the transcript by comparing the locus tags. If
the locus tag is not present, we try to retrieve the link between the
identifier of the transcript and the accession number of the protein from the
NCBI. If this is also not available, we use the product tag. The method used
for connecting the CDS and transcript is reported.
%\section{Annotation enrichment} \label{sec:enrichment}
\end{document}
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment