TAB Parser behaviour for blank entries in a column

[KeywanHP]

The below TAB parser snippet currently creates an "empty" concept even for blank entries in a given column. The desired behaviour would be to ignore blank entries.

    <concept id = "gene">
        <class>Gene</class>
        <data-source>ARAGWAS</data-source>
        <accession data-source="TAIR">
            <column index='1'/>
        </accession>
    </concept>

[marco-brandizi]

@KeywanHP, @jparsons2222, please someone test it with real data and close this if it works.

[KeywanHP]

This issue still exists. It is not related to a specific dataset but is a generic TAB parser enhancement request. For blank cells, the current logic seems to be to create a concept without any properties (ghost concept). The better logic would be to create concepts only for non-empty cells.

Do you agree @marco-brandizi @jparsons2222 ?

[marco-brandizi]

@KeywanHP, It should work as per the unit tests about it:

Sample file Config file test about missing values

As you can see,

• O43426 is created as a concept node, but not linked to anything, cause its row hasn't any other element
• Same for ENSG00000185345, since no protein accession is reported in the corresponding row
• ENSG00000184381 is created without any attribute (having the corresponding cells empty)

Another test (testBlankLinesFilter()) has been in place for long time, to verify that completely empty rows are ignored.

Shouldn't it behave this way? Do you have an example + how you expect it to be parsed? Are you using the latest version?

EDIT: the mentioned unit test had a typo (in conceptFinder), however I fixed it and it is still working without detecting errors for the behaviours I've expected so far.

[KeywanHP]

This is great. Exactly the behaviour that I would be expecting. We have been using knetbuilder 3.0. Should switch to the latest version now! @jparsons2222

[marco-brandizi]

Closing as fixed.