Posted by Richard Jacob (December 22, 2020)

How large a Mascot Server licence do I need?

One frequent question from new customers is how many CPUs do I need to licence for my Mascot Server? This is actually two questions. First, Mascot is licensed in CPU units, where 1 CPU is good for 4 cores. The computer can have more than one processor, and each processor can have more than 4 cores, but only the number of cores licensed will be used for the search. In general, doubling the number of licensed CPUs halves the search time if all other variables are kept the same. Doubling it again will quarter search time.

The major goal should be that the search engine can analyze the data quicker than the mass spectrometer(s) can acquire it. The second, key question is: how long does it take to search a data set? This depends on both the computer hardware and the search parameters. As you might imagine these are separate topics in their own right and have been covered in a series of posts titled “How long should a search take?” Part 1 covers “Data and search parameters” and part 2 “Hardware and configuration”.

The entry level is a 1 CPU licence, which is suitable for a lab that is just starting out with a single mass spectrometer, or a lab that needs to do its own analysis with data obtained from collaborators or core labs. If your lab is a service or core lab or moving beyond running your own samples to running the instrument 24/7, then a 2-4 CPU licence might be more suitable.

For long running projects where each sample has to be searched under the same conditions, you can set up Mascot Daemon to automatically search data as it comes off the mass spectrometer using a real time monitor. The search results from a night’s worth of runs will be waiting for you in the morning. Many labs can’t use a real time monitor task on a day to day basis as many samples are fractionated and need to be merged into a single search. Similarly, you may be running samples from multiple projects that require different search conditions. This means that 24 hours’ worth of data has to be searched in the 8 to 9 hours of a working day. Even if it took many days to acquire the data the end user might want the search results in hours.

Mascot Server licences with 5 or more CPUs are suitable for large core labs and labs with multiple instruments. These systems can really churn through data. You either need a computer equipped with two processors, each with many cores, or a cluster of commodity PCs.

Probably the best method to determine what size Mascot Server licence you need is to process some example files and see how long they take. These could be files obtained from a demo session with your potential new instrument or existing instrumentation or similar publicly available data sets from a repository like PRIDE. We are more than willing to analyze a test data set for you in order to obtain the best results and accurate search times. We recently ran a test analysis on a CPTAC project using the published search conditions. The data was searched against a humanRefSeq database included with the project and the Uniprot human proteome database. The searches were completed in 2 to 3 minutes per fraction using a 1 CPU Mascot Server licence. The entire data set for the project, seven days’ worth of acquisition time, could be searched in in less than 8 hours.

A change in sample type could mean that similar searches take a lot longer to process. If the samples were from an unknown bacterial species and had to be searched against all the bacterial sequences in NCBIprot or UniRef100, we might expect each fraction to take 20 to 30 minutes to analyze or 3 to 4 days for the whole project. A 2 or 3 CPU licence might be the better option.

What if the number of samples you are analyzing increases or you just need more speed? There is no financial penalty to upgrading the number of licensed CPUs at a later date – you just pay the difference in licensing cost. The upgrade can be as easy as registering a new product key. If you need to move Mascot to a more powerful computer, we have clear step by step instructions. On the other hand, we don’t recommend purchasing a licence that is far larger than you need as this will increase the cost of the initial computer system as well as annual maintenance.

Keywords: , , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *

*

HTML tags are not allowed.