ContentMine/Animal Models of Depression

This page documents the work pertaining to the ContentMine Fellowship awarded to Alexandra Bannach-Brown on the research topic of Animal Models of Depression.

Aims[edit | edit source]

To implement ContentMine tools to automate steps of the systematic review Animal Models of Depression. This research is being undertaken at CAMARADES, University of Edinburgh.

Full Text Retrieval[edit | edit source]

Aim: to use ContentMine tool 'getpapers' to retrieve Open Access full text articles.

Document Annotation & Classification[edit | edit source]

Aim: to use ContentMine tool 'ami' to annotate documents with existing ami dictionaries of gene and species data. Aim: to create regex dictionaries of common animal models of depression and of all known antidepressants.

Data Storage & Dissemination[edit | edit source]

Aim: to make information of the systematic review and meta-analysis of animal models of depression freely available and easily accessible.

Search[edit | edit source]

Currently working with the search string in the ContentMine tool 'getpapers' with this search string:

"getpapers -q "depressive disorder" OR "depression" OR "depressive behavior" OR "depressive behaviour" OR "dysthymia" OR "dysthymic" AND "animal" -o Depression -x -p -s"

This search retrieves 22003 hits via the EuPMC API (23rd Feb 2017).

I have had ongoing issues with the large number of studies retrieved using this search. The memory limit is often exceeded so the full text articles are not downloaded.

<--- Last few GCs --->
   89811 ms: Scavenge 1400.9 (1453.0) -> 1400.9 (1453.0) MB, 2.9 / 0 ms (+ 0.0 ms in 1 steps since last GC) [allocation failure] [incremental marking
delaying mark-sweep].
   90294 ms: Mark-sweep 1400.9 (1453.0) -> 1400.8 (1453.0) MB, 482.8 / 0 ms (+ 36.0 ms in 35 steps since start of marking, biggest step 3.0 ms) [last
resort gc].
   90783 ms: Mark-sweep 1400.8 (1453.0) -> 1400.7 (1453.0) MB, 489.0 / 0 ms [last resort gc].

<--- JS stacktrace --->
==== JS stack trace =========================================
Security context: 00000076666B4639 <JS Object>
    1: Join(aka Join) [native array.js:154] [pc=000003F3792B3CB8] (this=00000076666041B9 <undefined>,o=0000031BEAC3F019 <JS Array[40]>,v=40,C=0000025D
22C32769 <String[6]\: ,\n    >,B=000000766669BCA9 <JS Function ConvertToString (SharedFunctionInfo 0000007666658571)>)
    2: InnerArrayJoin(aka InnerArrayJoin) [native array.js:331] [pc=000003F3792B2B2A] (this=00000076666041B9 <undefined>,C=000...

FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - process out of memory

This issue has been partly solved by increasing memory allowance for the node programme from within 'Advanced System Settings' on a Windows PC. http://www.codexpedia.com/javascript/increasing-the-memory-limit-in-node-js/

After having increased the memory, I have made edits to my search string to limit the number of papers that are retrieved from the total search. This is done by adding

--limit 20000

to the search string. 20000 papers/records is currently the largest amount of papers I can download in a single command.

Annotation[edit | edit source]

Using the ContentMine tool to search through my corpus of depression documents to tag and annotate these document with relevant terms. These tags will be used to categorise documents/records into groups for more effective data extraction and meta-analysis, which are the next steps in the systematic review process.

Regex Dictionaries[edit | edit source]

Aims:

- To use regex dictionaries to tag articles relevant to the systematic review topic with instances of pertinent information.

- To create new regex dictionaries for relevant topics to the systematic review of animal models of depression.

I was fairly new to regex so first thing I did was do a google search. I found lots and lots of great tools and guides so I’ll summarise the points I learned here and useful resources. www.regular-expressions.info Is a great website that will teach you the basics of RegEx in one page! It also has a more in depth tutorials explaining each key concept and expression.

Great tools to test out your regular expressions on an example text:

RegExr: http://regexr.com/

RegEx101: https://regex101.com/

Animal Models of Depression[edit | edit source]

The regular expression dictionary for animal models of depression was created by locating suitable resources to ensure the list was as extensive as possible. I searched for reviews and pinnacle papers in the field and started my list (the references I used can be found here).

After putting together my regex list of models of depression, I put it into a format that 'ami' can read, a .xml file. ContentMine had great documentation about how to go about this, and I had no issues. https://github.com/ContentMine/workshop-resources/blob/master/software-tutorials/ami/README.md#ami2-regex

My animal models of depression regex can be found here.

Simply add your regex.xml file to your project directory and execute using the ami2-regex command. View your instances in the results folder of each record within the CTree.

While my list of models of depression is by no means exhaustive, it is meant as a starting point upon which I can group my papers, and ease the annotation stage by using some automation tools.

Antidepressants[edit | edit source]

The regular expression dictionary for antidepressants was created from the Wikipedia page 'List of Antidepressants'. This page was accessed on 10-01-2017 when the data was extracted. On top of this list, I visited the wikipedia page of each drug listed, where available, to note down any additional trade names or chemical names of drugs in order to for the regex to pick up on any instances of the drug in the literature.

Using the same strategy as for the Animal Models of Depression regex dictionary, these words converted into a format the 'ami' can read, using ContentMine's documentation.

My antidepressant regex dictionary can be found here.