Hello,
I am trying to extract a value from the XML below based on the value of the previous record before the current record. Basically, if the Textseq-id_accession of the previous record starts-with 'NM_' I want to to select that value as the 2nd "column" in my output. I have tried preceding and ancestor but the either take me to the top of the document or shows the NP_ value rather than the NM_.

<?xml version="1.0"?>
<Bioseq-set>
  <Bioseq-set_seq-set>
    <Seq-entry>
      <Seq-entry_set>
        <Bioseq-set>
          <Bioseq-set_seq-set>
            <Seq-entry>
              <Seq-entry_seq>
                <Bioseq>
                  <Bioseq_id>
                    <Seq-id>
                      <Seq-id_other>
                        <Textseq-id>
                          <Textseq-id_accession>NM_201656</Textseq-id_accession>
                          <Textseq-id_version>1</Textseq-id_version>
                        </Textseq-id>
                      </Seq-id_other>
                    </Seq-id>
                  </Bioseq_id>
                </Bioseq>
              </Seq-entry_seq>
            </Seq-entry>
            <Seq-entry>
              <Seq-entry_seq>
                <Bioseq>
                  <Bioseq_id>
                    <Seq-id>
                      <Seq-id_other>
                        <Textseq-id>
                          <Textseq-id_accession>NP_973385</Textseq-id_accession>
                          <Textseq-id_version>1</Textseq-id_version>
                        </Textseq-id>
                      </Seq-id_other>
                    </Seq-id>
                  </Bioseq_id>
                </Bioseq>
              </Seq-entry_seq>
            </Seq-entry>
          </Bioseq-set_seq-set>
        </Bioseq-set>
      </Seq-entry_set>
    </Seq-entry>
  </Bioseq-set_seq-set>
</Bioseq-set>

My current XSL looks like this:
<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
        <xsl:output method="text" encoding="ISO-8859-1" />


        <xsl:template match="/">
            <xsl:apply-templates
                select=".//Bioseq-set/Bioseq-set_seq-set/Seq-entry/Seq-entry_set/Bioseq-set/Bioseq-set_seq-set/Seq-entry"
                mode="DocRoot" />
        </xsl:template>


        <xsl:template match="Seq-entry" mode="DocRoot">
            <xsl:for-each select="Seq-entry_seq/Bioseq/Bioseq_id/Seq-id/Seq-id_other/Textseq-id">
                <xsl:if test="starts-with(Textseq-id_accession,'NP_')">

                    <xsl:value-of
                        select="Textseq-id_accession"/>
                    <xsl:text disable-output-escaping="yes">&lt;&gt;</xsl:text>

                    <xsl:value-of select="ancestor::Seq-entry_seq/Bioseq/Bioseq_id/Seq-id/Seq-id_other/Textseq-id/Textseq-id_accession"/>
<xsl:text disable-output-escaping="yes">&lt;&gt;
</xsl:text>
                </xsl:if>
            </xsl:for-each>
        </xsl:template>

</xsl:stylesheet>

Output currently is:
NP_973385<>NP_973385<>

Should be
NP_973385<>NM_201656<>

And if the previous Textseq-id_accession was not like NM_, then a blank should be returned.

I have looked at Jeni's pages, but I am a bit baffled by all the grouping as I am not sure if that is what I want (looking at some other posts here they seem to want something slightly different).

First, I'm wondering when I read:

<xsl:template match="/">
            <xsl:apply-templates
                select=".//Bioseq-set/

whether you really want to read all the descendant Bioseq-set elements here, or only the one at the top level? If you only want the top-level, then searching the whole document for it is very inefficient.

Secondly, when you're using output method="text" then you don't need disable-output-escaping, because the text output method never does any escaping in the first place.

Looking at this expression:

<xsl:value-of select="ancestor::Seq-entry_seq/Bioseq/Bioseq_id/Seq-id/Seq-id_other/Textseq-id/Textseq-id_accession"/>

when you're positioned in the NP_973385 entry the only ancestor Seq-entry_seq is within the second of the two Seq_entry elements, so I can't see why you imagine it will select something within the first Seq_entry. From your description, I would think the right path expression is

<xsl:value-of select="ancestor::Seq-entry/preceding-sibling::Seq-entry[1]/Seq-entry_seq/Bioseq/Bioseq_id/Seq-id/Seq-id_other/Textseq-id/Textseq-id_accession"/>

or perhaps more simply

<xsl:value-of select="preceding::Textseq-id_accession[1]"/>

Michael Kay

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Thanks very much Michael, this was very helpful. I understand my mistake with the ancestor now. For the template macthing the root. Do you mean there is a syntax issue with the .// or is there something else I am not understanding. The Bioseq-set is the top level of the document - I was just wanting to start from the top of the document and move down to the Seq-entry records.

Hello All again. Here is my next problem, similar to above. I have attached a sample of the xml. The XML is split into Seq-entry records. For those Seq-entry records whose "Textseq-id_accession" starts-with 'NP_', I pull out some extra information. Unfortunately, it seems that I need the "GeneID" value of the NP record. The GeneID is stored in another Seq-entry record. The GeneID for each NP record is held in just one record under the
<Bioseq-set>
  <Bioseq-set_seq-set>
    <Seq-entry>
      <Seq-entry_set>
        <Bioseq-set>
          <Bioseq-set_seq-set>
record-set (the first Seq-entry record here has a "Textseq-id_accession" that does not start with NP). For each GeneID there is a link to the NP record in the "Gene-ref_locus-tag" tag. This tag in the gene record matches the
           <Seq-entry>
              <Seq-entry_seq>
                <Bioseq>
                  <Bioseq_id>
                    <Seq-id>
                      <Seq-id_general>
                        <Dbtag>
                          <Dbtag_tag>
                            <Object-id>
                              <Object-id_str>
tag of the NP record. How can I go through each NP record and extract the GeneID value from the 1st record based on the matching of the Gene-ref_locus-tag and the Object-id_str?

<?xml version="1.0" encoding="UTF-8"?>
<Bioseq-set>
  <Bioseq-set_seq-set>
    <Seq-entry>
      <Seq-entry_set>
        <Bioseq-set>
          <Bioseq-set_seq-set>
            <Seq-entry>
              <Seq-entry_seq>
                <Bioseq>
                  <Bioseq_id>
                    <Seq-id>
                      <Seq-id_other>
                        <Textseq-id>
                          <Textseq-id_accession>NC_005054</Textseq-id_accession>
                          <Textseq-id_version>1</Textseq-id_version>
                        </Textseq-id>
                      </Seq-id_other>
                    </Seq-id>
                  </Bioseq_id>
                  <Bioseq_annot>
                    <Seq-annot>
                      <Seq-annot_data>
                        <Seq-annot_data_ftable>
                          <Seq-feat>
                            <Seq-feat_data>
                                <Gene-ref>
                                  <Gene-ref_locus>repA</Gene-ref_locus>
                                  <Gene-ref_locus-tag>VRA0001</Gene-ref_locus-tag>
                                </Gene-ref>
                            </Seq-feat_data>
                            <Seq-feat_dbxref>
                              <Dbtag>
                                <Dbtag_db>GeneID</Dbtag_db>
                                <Dbtag_tag>
                                  <Object-id>
                                    <Object-id_id>2598373</Object-id_id>
                                  </Object-id>
                                </Dbtag_tag>
                              </Dbtag>
                            </Seq-feat_dbxref>
                          </Seq-feat>
                          <Seq-feat>
                            <Seq-feat_data>
                                <Gene-ref>
                                  <Gene-ref_locus-tag>VRA0002</Gene-ref_locus-tag>
                                </Gene-ref>
                            </Seq-feat_data>
                            <Seq-feat_dbxref>
                              <Dbtag>
                                <Dbtag_db>GeneID</Dbtag_db>
                                <Dbtag_tag>
                                  <Object-id>
                                    <Object-id_id>2598374</Object-id_id>
                                  </Object-id>
                                </Dbtag_tag>
                              </Dbtag>
                            </Seq-feat_dbxref>
                          </Seq-feat>
                        </Seq-annot_data_ftable>
                      </Seq-annot_data>
                    </Seq-annot>
                  </Bioseq_annot>
                </Bioseq>
              </Seq-entry_seq>
            </Seq-entry>
            <Seq-entry>
              <Seq-entry_seq>
                <Bioseq>
                  <Bioseq_id>
                    <Seq-id>
                      <Seq-id_general>
                        <Dbtag>
                          <Dbtag_tag>
                            <Object-id>
                              <Object-id_str>VRA0001</Object-id_str>
                            </Object-id>
                          </Dbtag_tag>
                        </Dbtag>
                      </Seq-id_general>
                    </Seq-id>
                    <Seq-id>
                      <Seq-id_other>
                        <Textseq-id>
                          <Textseq-id_accession>NP_877980</Textseq-id_accession>
                          <Textseq-id_version>1</Textseq-id_version>
                        </Textseq-id>
                      </Seq-id_other>
                    </Seq-id>
                  </Bioseq_id>
                </Bioseq>
              </Seq-entry_seq>
            </Seq-entry>
            <Seq-entry>
                <Seq-entry_seq>
                    <Bioseq>
                        <Bioseq_id>
                            <Seq-id>
                                <Seq-id_general>
                                    <Dbtag>
                                        <Dbtag_tag>
                                            <Object-id>
                                                <Object-id_str>VRA0002</Object-id_str>
                                            </Object-id>
                                        </Dbtag_tag>
                                    </Dbtag>
                                </Seq-id_general>
                            </Seq-id>
                            <Seq-id>
                                <Seq-id_other>
                                    <Textseq-id>
                                        <Textseq-id_accession>NP_877981</Textseq-id_accession>
                                        <Textseq-id_version>1</Textseq-id_version>
                                    </Textseq-id>
                                </Seq-id_other>
                            </Seq-id>
                        </Bioseq_id>
                    </Bioseq>
                </Seq-entry_seq>
            </Seq-entry>
          </Bioseq-set_seq-set>
        </Bioseq-set>
      </Seq-entry_set>
    </Seq-entry>
  </Bioseq-set_seq-set>
</Bioseq-set>

The XSL I have so far is :

<xsl:transform xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
<xsl:output method="text" encoding="ISO-8859-1" />


   <xsl:template match="/">
      <xsl:apply-templates
          select=".//Bioseq-set/Bioseq-set_seq-set/Seq-entry/Seq-entry_set/Bioseq-set/Bioseq-set_seq-set/Seq-entry"
          mode="DocRoot" />
   </xsl:template>


    <xsl:template match="Seq-entry" mode="DocRoot">
        <xsl:for-each select="Seq-entry_seq/Bioseq/Bioseq_id/Seq-id/Seq-id_other/Textseq-id">
            <xsl:if test="starts-with(Textseq-id_accession,'NP_')">

                <xsl:value-of select="Textseq-id_accession"/>
                        <xsl:text>&lt;&gt;</xsl:text>


<xsl:text >
</xsl:text>
            </xsl:if>
        </xsl:for-each>
    </xsl:template>

 </xsl:transform>

Output would be

NP_877980<>2598373
NP_877981<>2598374

>Do you mean there is a syntax issue with the .//

No, I didn't suggest it was a syntax error, I just asked why you were using .//x when it appears that x would suffice, given that .//x is searching the whole document while x finds the required element immediately. Of course I haven't the faintest idea of the semantics of your (rather peculiar) XML, so I've no idea whether your code is right or wrong.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Thanks for your clarification. I had just used the .// as I had seen that used in another XSL. I am having to learn this on the fly so I have taken your advice on board and have changed the XSL appropriately.
Many thanks
Steve.