Condel in the world of web services

We have found two major drawbacks to computational approaches that assess the outcome of non-synonymous SNVs. The first is that there are now too many of them, based on different principles –some using phenomenological scoring functions, some that take advantage of machine learning strategies– but few efforts to make them comparable and integrate their outputs into a consensus that may be readily interpreted and employed by researchers. Our group and others have taken a first step towards the development of approaches to make such consensus possible. We recently developed and presented an approach to compute a Consensus Deleteriousness score (condel) of these computational tools and tested it with five of them.

Now the second problem with most existing tools is that they are not currently available for download or as web services to incorporate to New Generation Sequencing workflows. Many of them are just available through webservers, where the user must copy-and-paste their SNVs or upload a limited number of them, so that the methods are run on the server side and their results are retrieved later in HTML or flat file formats. So, it is difficult for groups or projects generating large catalogues of non-synonymous SNVs (think, for instance in 1000genomes, or the ICGC) to rapidly assess their potential impact on protein function. In order to be prepared for the rapid analysis of millions of variants resulting from such large sequencing efforts it is necessary to provide tools that may be easily downloaded and installed by the user or –even better– web services that may be queried remotely. This second option relieves the user from the laborious effort of installing the tool on their side, but still allows them to incorporate them to their workflows. One example of such a web service is the Ensembl-variation API which provides access to pre-computed SIFT and PolyPhen scores of all possible SNVs in the human proteome, as well as their condel integration, presented in a previous post of this blog.

Now, we have developed our own condel web service (condel-server) which allows the fast and easy integration of the individual scores of SIFT, PolyPhen and MutationAssessor for SNVs. Condel-server is implemented in java, as a modular application designed using well-known OO design patterns. It may be queried with SNVs both in genomic coordinates and in protein coordinates. It uses the Ensembl-variation API to retrieve SIFT and Polyphen scores and the uses the MutationAssessor webAPI to retrieve its score. Condel server may be easily incorporated to any SNV analysis workflow employing two simple commands. First, a tab separated file containing the SNVs in the formats specified here may be uploaded with the command:

curl -X PUT -T [input_file] http://bg.upf.edu/condel-server/taskService

which returns the id assigned to the task created by condel server to respond to the user’s request. Then, a tab-separated file with the output may be retrieved using:

curl -X GET http://bg.upf.edu/condel-server/taskService/[task_id]

Nevertheless, condel server may be queried also using the traditional web interface where SNVs may be either copied-and-pasted or uploaded in a tab separated file. This is what the web interface looks like:

This interface offers a lot of possibilities to manage the user’s tasks and filter their results using different criteria. A simple tutorial of its usage may be found here, along with a through help section that guides the user through the server’s capabilities. So, try condel server, we are happy to hear any feedback you have for us.