Hi David,

Please find below some further comments.

On 18 May 2010 02:45, David Gallagher <gallagher.da at gmail.com> wrote:
> Hi Vedrin,
>
> I ran a quick test with ToxPredict and have the following comments on the
> user interface.  My first test involved entering my own structure to make a
> toxicity prediction.
>
> First impressions: the interface seems to have progressed well since I last
> looked and is much more robust with detailed explanations to help the first
> time user, and with no problems or 404 errors etc. noted.  Obviously, the
> results/reporting function still needs to be addressed, so I have ignored
> that for now.

This is work in progress.

> 2.  I entered the SMILES string for my structure into the box, then clicked
> "Next".  The screen refreshed and my entry was deleted.  I saw an error
> message "We did not find any matching entries for the search you performed
> in the OpenTox database. Please try again.", however, I was not expecting to
> search any database, but simply wanted ToxPredict to make a prediction on my
> structure.

Well, in the contextual help on this screen it is clearly stated what
the functionalities provided by the Search and Draw tabs are, quote:

• Search: You can enter any free text, including but not limited to
CAS registry number, EINECS number, SMILES, InChI, chemical name, or
any other keyword; ToxPredict will automatically recognize the
intended search type and return zero, one or more hits, depending on
the search type, the information currently available in the OpenTox
database and the selected preferred number of hits.
• Draw: You can draw a chemical structure in the built-in editor and
perform exact structure or substructure search, or structural
similarity based search; ToxPredict will return zero, one or more
hits, depending on the search type, the information currently
available in the OpenTox database and the selected preferred number of
hits.

Moreover, the same help section also indicates that if you would like
to import your own structures, you should use the Upload tab, quote:

•Upload: You can upload your own dataset(s); ToxPredict accepts SDF,
MOL, CSV and RDF input file formats, among many others.

>  I checked but could see how to do this.  Am I missing something?

Do you think that the Help text is not clear? Would you advise some
changes in it?

> 3.  When I sketched a structure using the "Draw" tool, I got the same error
> as in (2) above.

This is expected behaviour (see comments in point (2) above).

I would like to mention that we could easily broaden the semantics of
Search and/or Draw tabs to include data upload, in addition to
database search. There are various ways to do this, e.g:

1) perform a database search and if no hits are returned AND the user
has entered a valid SMILES/drawing, transparently accept the user
supplied structure for prediction, in the same way as this would be
achieved through the Upload tab;

2) same as (1), but asking the user to confirm that he agrees to
import this structure in the database;

Please also consider that some users might not be happy to import
their own structures in the OT database for various reasons., even if
Authentication and Authorization was already in place. Hence, the need
to clearly separate Search from Upload (at least in our opinion) ...

> 4.  Uploading a Mol file, this worked as expected.
>
> Some minor user-interface issues which could improve the look and feel of
> ToxPredict:
>
> 1. "NEXT" button
> For most users of the Roman alphabet, the eye naturally scans from top left
> to bottom right of a page so, after entering any required data, the last
> place one intuitively looks for the "Next" step  is usually the bottom
> right.  With the current design, the "NEXT" button can appear at the top
> right or sometimes the top left (depending on page width).  The bottom right
> would be a more intuitive place to place the "NEXT" button.

Agreed. Suggested fix: make the Navigation bar appear both on top and
at the bottom of the screen. Do you think this would solve the issue?

> 2. Instructions
> The text instructions, although very helpful, are unnecessarily long.  Some
> of the text is in triplicate.  I would suggest only succinct explicit
> instructions to be displayed, with a "more info" link if necessary.

Well, obviously in addition to being long, the text is not clear
enough, considering the above mentioned problems you've encountered
with the Search and Draw tabs. Could you suggest better (shorter and
more concise) contextual help texts for each of the screens? In
addition, we could hide the Help text in a +Help menu item, which
would be expanded only if the user clicks on it -- would this be more
convenient in your opinion?

> 3. LOGOs
> The OpenTox and Ambit logos are too large and draw the user's attention away
> from the functionality.  They would be better placed in the centre at the
> bottom of the page, and made smaller.

Please specify the dimensions of the logos that you deem acceptable
and we would try to apply them.

> 4.  Display results
> The "Display results' seems to be an unnecessary manual step, the results
> should be displayed automatically after the calculations are completed.

Disagree. At the Estimate page, the users should be able to check
which calculations have been completed successfully before going to
the Results page. In case that there are some errors, the users should
be able to check the error details at the Estimate page as well,
instead of being automatically transferred to the Results page.
Depending on the errors, they might prefer to run the prediction again
or select different models,etc ..., instead of going to the results
page.

> 5.  Data entry
> Ideally, the data entry (i.e. Search, Draw, or Upload) should be independent
> of the subsequent Database Search and Predict functions.  Currently, only
> "Uploaded" structures can be used for predictions, and not "Drawn"
> structures.

The statement in your last sentence is wrong. Any structure (or even
set of structures), that have been found in the OpenTox database to
satisfy the search criteria that the user has defined either in the
Search or Draw tabs can be used for predictions. The OT database
contains a superset of the full list of REACH-relevant substances
(currently 150369 chemicals). Of course, substances that are not
included in the OT database could also be submitted for predictions
through the Upload tab. As discussed already above, please specify
whether you think that the semantics of Search/Draw should be enriched
to include Upload as well, in addition to this same functionality,
provided by the Upload tab.

The structure you've uploaded doesn't happen to exist in OT database
for some reason. A closer look reveals quite a few similar structures
though:

http://apps.ideaconsult.net:8080/ambit2/query/similarity?type=url&search=http%3A%2F%2Fapps.ideaconsult.net%3A8080%2Fambit2%2Fcompound%2F150368%2Fconformer%2F435224&threshold=0.65&max=100

Yours is the last one in the list. If you had selected a similarity
(or substructure) search in the Draw tab (instead of exact structure
match you've obviously performed, which is the default), then your
query would have returned some of these chemicals, depending on the
selected Tanimoto coefficient and desired number of hits. Again, any
suggestions for better wording of the Help text, explaining these
functionalities would be most welcome.

> This section seems to function inconsistently at present.

Well, it depends on what you perceive as consistent. Obviously our
perception is different from yours, but we would be happy to implement
yours, provided that you specify it in more details ;-)

> 6.  Standardisation on terminology
> For consistency, it may be better the change "Step 4, Estimate" to "Step 4,
> Predict", as this would then match the title "ToxPredict".  Otherwise, the
> question arises, what is the difference between Estimate and Predict?

Agreed.

> 7.  Tag line
> A tag line such as "Toxicity predictions, just enter your structure..."
> under the title would succinctly explain the function of the page.  "OpenTox
> Demo application" could be repositioned at the bottom of the page.

Agreed.

> 8.  Disclamier
> There should be a disclaimer at the bottom of the page (in small print)
> explaining that computational chemistry predictions should not be relied
> upon alone, but used as part of a weight of evidence, etc., etc......

Agreed. Could you suggest the text for this Disclaimer?

> Please let me know if you need any further explanation on any of the points
> I have mentioned.
>
> All in all, ToxPredict seems to be progressing well, good work!
>
> Regards,
> David

Last, but not least -- it would be also nice to hear some other
opinions on the controversial topics -- anyone?

Kind regards,
Vedrin