Pipeline is Aspen Biosciences premiere application for managing drug discovery projects. It started as a simple dashboard for visualising the drug discovery pipeline, and has grown into what some project scientists characterise as “JIRA for Pharmaceutical Projects”. That growth has been driven largely by conversations with scientists who’ve found themselves in the position of having to manage drug discovery projects and needing a tool that was both lightweight and user-friendly.
In the previous series of blog posts, we talked about Design Thinking and how it can be used to create research-oriented software to support drug discovery. In this series of blog posts, we’ll take a look at some of the stories that scientists shared with us, and show you examples of how we translated conversations into new functionality for Pipeline.
A Scientist Walks Into A Bar
Here in San Diego we are blessed with a number of organisations designed to help the entrepreneurial scientist translate their ideas into businesses. One of these organisations is the San Diego Entrepreneur Exchange (SDEE). Every month they host a Happy Hour meeting at a local microbrewery. It’s a great way to unwind, and network, and for us it’s a great way to meet scientists and listen to their stories.
At one such gathering a few years ago, I ran into a Director of Discovery Biology at a small drug discovery company (let’s call her Jane — not her real name, gender or title). I introduced myself, and mentioned that I write software for drug discovery and since I’ve always been interested in the process and tools that companies use to select their targets I wanted to learn more about how they selected targets at her company.
Jane’s company, is a small drug discovery company with about 50 scientists on staff that had been in business for over 14 years and had successfully partnered with larger companies to get their drugs to market before finally being acquired. Even after the acquisition though, the company still continue to function as though it were a small biotech. So I was interested to hear how they managed the target selection process in this blended environment.
I began by asking her how the target selection process had changed over the years. “What tools are you using now to manage the drug target selection process?”
“A few years ago we bought a license for a product by a local company. [I’ve omitted the name to protect the innocent]. The product has changed hands over the years and it’s become almost a kitchen sink application.”
“One of the things they tout, is that ‘we do all the homework for you, so you don’t have to. We go to all the conferences. We research all of the targets and the pathways, we curate all of this information for you and save you all that time and money. And for all of that we charge $50,000 a year’.”
“And while I appreciate the amount of work that goes into curating all of that information, it’s a little disingenuous to imply that we can simply delegate that work to them and hope that they’ve done it properly. We still need to research those targets ourselves, we do a deep dive into the target and the indication. We look at disease biology, we look at clinically relevant variants, we look at a half-dozen pathway databases. We look at specialist databases like COSMIC and TCGA (The Cancer Genome Atlas) for specific cancer indications. We look to see what drugs are in development for the target. We do literature searches in PubMed and read lots of articles. We go to conferences, we create internal target profiles, and presentations. But $50K is still pretty steep for something that only one person might use 2-3 times a year, and for work that we’re going to do anyway. $50K would keep us in reagents for a few months. We could buy another microscope — something we’d use every day.”
“So what were you doing before? How were you organising the information,” I asked.
“We used PowerPoint and SharePoint for a while. We’d create a presentation template to help us pull the information together. You still have to go to all of the sites, review the information and curate it. But at least the information was curated in a single consistent format. ”
“What made you switch from SharePoint,” I asked.
“With SharePoint, only one person at a time could work on the target profile. We have a number of people in our group and we needed to divide the work between us. But splitting the work up between the members of the team had its consequences as well. When we were first getting started, we realised that we weren’t doing consistent searches in PubMed. Now we teach target reviewers to do searches consistently. They can still create their own custom searches, but we want to make sure that there’s at least a minimal set of searches being done, and that the search criteria are consistent from project-to-project. ”
“But the real problem with SharePoint was that it didn’t understand biology. We wanted to know the Gene Ontology terms for the target. We want to know about the domains that the target has. We want to know which pathways a target plays a role in. We wanted to be able to look across our portfolio at our projects see if there were similarities between projects that we’d already worked on.”
“And now? With your new application? Do you feel that those issues were addressed” I asked.
“To be honest it feels like we’ve traded one set of problems for another. We see these curated views, that aren’t editable. We can’t emphasize that this subset of the available information is pertinent to our project, and this other set isn’t. We need a better alternative.”
Translating The Conversation Into Code
One of the key requirements that Jane shared with us was the need to provide a single point of entry for a project. There needed to be a way to import target-related data from multiple sources to create a single curated view of the target. To do this, we implemented a search tool that lets the user easily import data from UniProt, PharmGKB, EntrezGene, PubMed, Protein Data Bank, ChEMBL and many other sources with a single query.
Documents and Presentations
In Jane’s company they use Google Docs to collaborate on documents, presentations and spreadsheets. They made the switch to Google Docs after a particularly tense day when a team of researchers were trying to prepare a presentation highlighting results from their latest project. The problem was they were attempting to collaborate on the same presentation by sending copies of it back and forth by email. Versions of the document were getting muddled, slides which had been added, disappeared in the next version, back and forth it went and tensions began to rise.
Cloud-based office applications like Google Docs and Office 365 have become more ubiquitous in pharmaceutical companies over the years but in many cases users aren’t made aware that they can actually simultaneously edit a document. At Aspen, Dimitri and I use this feature to collaborate on proposals and presentations while chatting with colleagues in San Francisco or Boston.
In both Google Docs and Office 365, users store the URL for the document in Pipeline so that they can easily get to the information they need. Document-level security is managed by either Google or Microsoft.
In scenarios where a document is authored by a single user, the document can be uploaded into a private Google Drive that is part of Pipeline’s backend infrastructure. In cases where the customer uses a document management system for sharing, signing and managing document workflows, users can publish the URL for the document in Pipeline along with document metadata such as a title, description and tags used to classify the document.
Pathways, PubMed and More
A significant portion of Jane’s time is spent in portfolio management. In many companies with large drug development organisations, the focus of portfolio management is financial, competitive and logistical. In drug discovery, the focus of portfolio management is primarily scientific and can be boiled down to two questions – what is the role of this target with respect to the indication, and how amenable is it to therapeutic intervention.
To answer these questions requires information from a variety of different sources to be integrated.
- EntrezGene is an NCBI database of genes and their sequences, sequence annotations and functions.
- ChEMBL is a database of targets and drugs created by the EMBL. Information from this database tells researchers about the drugs currently available or in development for a given target
- GeneRIF – these References Into Function provided by NCBI’s EntrezGene database, are a collection of papers that describe the role of a gene with respect to a variety of different indications.
- UniProt – is a database that contains information about proteins and their functions.
- OMIM – is the Online Medelian Inheritance In Man database contains information about genes and their roles in various indications.
- Protein Data Bank (PDB) – is a database of protein structural information that provides researchers with 3D models for drug targets.
- Gene Ontology information which describes the biological function, molecular function, and cellular localisation of genes and their protein products.
- PharmGKB is the Pharmacogenomics Knowledge Base which contains information about the target and how existing variations affect the efficacy of known drugs
- PubMed – is a massive database of millions of academic papers
- Pathway information from sources like KEGG, Pathway Commons, WikiPathways and the NCI’s Pathway Interaction Database are used to help researchers understand the role of the target with regard to biological networks.
By providing all of this information in a single place that was easily curateable by project scientists, Jane’s team was able to create the kind of collaborative working environment that she had been looking for. And the Pipeline Stats View let her look at projects across her entire portfolio for areas of commonality that they might be able to exploit.
Read more of the Pipeline Stories series.
Need Help Getting Started
To find out more about what Pipeline, and what we can do for your research organisation, contact us.