[Bioperl-pipeline] Document on input iohandler

[Xiao Juguang]

Hi guys,

After talking with Shawn, the biopipe chief and chef, we need more
illustration on one of the trickiest part in biopie, input iohandler.

Q: Why do we not need to specify input iohandler for analysis, in the XML
file?

A: In the template/genome_annotation_test.xml, we did not see any
'input_iohandler' tag inside the 'analysis' tag, while you find each
'analysis', except data_monger, has one 'output_iohandler' tag. Maybe your
question will raise, how does the analysis know which iohandler it will
apply to fetch the input? What the analysis will directly know from the XML
is which input it will consume, and the input table store which iohandler
will be called to fetch the input.

Q: How to use converter in the pipeline?

A: An converter is attached to an iohandler. If the iohandler's type is
INPUT, then the iohandler will fetch the data first then converter the data
from one format to another.If the iohandlers' type is OUTPUT, it will
converter the data first and then store. The converter is specified by the
combination of analysis and iohandler. That means many analysis can share
the same iohandler but with different converter.

if you want to add a converter into an iohandler, you should write XML in
this way,

<analaysis id="4">
       ...
    <output_iohandler id="3">
        <converter id="1" module="feature_pair_bio2ens"/>
    </output_iohandler>
</analysis>

we do not support input conversion so far.

Juguang.