RFC009

Title: DASBED Format
Author: Jim Kent
Version: 1
Date: 18 August 2001


I'd like to propose a format that is closer to what's native for

the UCSC browser. I think it's more flexible, a bit

better defined, and definitely denser than DASGFF.

 

Here's the DTD with comments to the right

 

<!ELEMENT DASBED (TEXT? SEGMENT*)>      Top level data structure
<!ATTLIST DASBED id CDATA #REQUIRED>         Some database ID.

<!ATTLIST DASBED label CDATA #IMPLIED>       One or two word label. Defaults to id.
<!ATTLIST DASBED description CDATA #IMPLIED> One line description. Defaults to label.
<!ATTLIST DASBED version CDATA "1.0">        Version must be floating point number
<!ATTLIST DASBED author CDATA "anonymous">   Overall author list.

<!ATTLIST DASBED href CDATA #IMPLIED>        Top level URL.  See below for details

 

<!ELEMENT TEXT (#PCDATA)>             A tag around a possibly long description.

 

<!ELEMENT SEGMENT (TEXT? TRACK*)>       Encloses annotations around a segment of genome

<!ATTLIST SEGMENT id CDATA #REQUIRED>     Sequence id (as in DASGFF's SEGMENT)

<!ATTLIST SEGMENT start CDATA #REQUIRED>   Start coordinate (1 based)
<!ATTLIST SEGMENT end CDATA #REQUIRED>     End coordinate (inclusive)
<!ATTLIST SEGMENT version CDATA #REQUIRED> Goes with sequence id.
<!ATTLIST SEGMENT label CDATA #IMPLIED>    One or two word label. Defaults to seq:start-end
<!ATTLIST SEGMENT description CDATA #IMPLIED> One line description.  Defaults to label.

 

<!ELEMENT TRACK (TEXT? I*)>             A track of data over a specific segment
<!ATTLIST TRACK id CDATA #REQUIRED>          Some database ID.

<!ATTLIST TRACK label CDATA #REQUIRED>       One or two word label.  Defaults to id.
<!ATTLIST TRACK description CDATA #REQUIRED> One line description. Defaults to label.
<!ATTLIST TRACK version CDATA #IMPLIED>      Floating point number.
<!ATTLIST TRACK author CDATA #IMPLIED>       Track author.  Defaults to DASBED URL.
<!ATTLIST TRACK href CDATA #IMPLIED>         Track level URL. Defaults to DASBED URL.

 

<!ELEMENT I (TEXT? B*)>             An item on a track, may contain multiple blocks.
<!ATTLIST I start CDATA #REQUIRED>       Start of first block relative to segment start.
<!ATTLIST I end CDATA #REQUIRED>         End of last block relative to segment start.

<!ATTLIST I id CDATA #REQUIRED>          Some database ID.
<!ATTLIST I label CDATA #IMPLIED>        One or two word label.  Defaults to id.
<!ATTLIST I description CDATA #IMPLIED>  One line description.  Defaults to label.
<!ATTLIST I version CDATA #IMPLIED>      Optional version here mostly for curated items.
<!ATTLIST I author CDATA #IMPLIED>       Optional author mostly for curated items.
<!ATTLIST I href CDATA #IMPLIED>         Item level URL.  Defaults to TRACK URL.
<!ATTLIST I strand CDATA #IMPLIED>       +, -, or not even present.
<!ATTLIST I score CDATA #IMPLIED>        A number between 0 and 1000.
<!ATTLIST I thickStart CDATA #IMPLIED>   CDS start for genes.
<!ATTLIST I thickEnd CDATA #IMPLIED>     CDS end for genes.

<!ATTLIST I qId CDATA #IMPLIED>          ID of query sequence in an alignment.

 

<!ELEMENT B>                        A block.  Can hold alignment info.  Deliberately small.
<!ATTLIST B start CDATA #REQUIRED>       Start in sequence relative to item start.
<!ATTLIST B end CDATA #REQUIRED>         End in sequence relative to item start.
<!ATTLIST B score CDATA #IMPLIED>        Score between 0 and 1000.
<!ATTLIST B qAli CDATA #IMPLIED>         Query side of alignment including '-' chars.
<!ATTLIST B tAli CDATA #IMPLIED>         Target (in segment) side of alignment.

<!ATTLIST B qStart CDATA #IMPLIED>       Query start coordinate.

<!ATTLIST B qEnd CDATA #IMPLIED>         Query end coordinate.

 

An example:

 

<DASBED id="hg7" label="April 2001" description="UCSC April 2001" author="various">

   <TEXT>This is data from the UCSC genome browser/database

         based on the April 2001 assembly on the Wash U map.</TEXT>

   <SEGMENT id="chr22" start="13000000" end="13020000" version="1.0">

       <TRACK id="exoFish" label="Exofish" description="Exofish ecores"

       href="http://www.genoscope.fr/cgi-bin/lookup?$I">

          <I start="1228" end="1399" id="e10223" score="880" ->

          <I start="10533" end="10988" id="e10224" score="800" ->

          <I start="13999" end="18800" id="e10225" score="999">

              <TEXT>I'm dubious about this one. -jsbach@genoscope.fr</TEXT>

          </I>

       </TRACK>

       <TRACK id="blatMouse" label="Mouse Alignments" author="WJ Kent">

          <TEXT>These Mouse/Human alignments were generated with BLAT using

                a 7 of 8 seed at the amino-acid level.  The mouse sequence

                was generated by the Mouse Sequencing Consortium and

                assembled with Arachne. </TEXT>

          <I start="1038" end="1077" id="ti|18293" score="901" strand="+">

            <B start="1" end="10" tAli="ATAATG--GTTG" qAli="ATAACGAAGTTG" qStart="254" -> 
            <B start="28" end="39" tAli="ATTATGGCTG" qAli="ATTACGGCTG" qStart="254" ->

          </I>

       </TRACK>

    </SEGMENT>

</DASBED>

 

The href's can have the following variables embedded in them:

   $I - replaced with the item ID.

   $T - replaced with the track ID.

   $G - replaced with the global bed ID.

   $S - replaced with sequence ID

   $A - replaced with start of item.

   $Z - replaced with end of item.

These will be substituted before any linking takes place.   Most of the

time it will be possible to specify the URL at the track or even global level.