2/19/2008 MaineTech BioSIG meeting postponed

February 19th, 2009
Tonight’s (2/19/2009) presentation at the MaineTech Bioinformatics Special Interest Group meeting has been postponed due to weather. It has been rescheduled for March 19th.

Biological Signal Processing (BSP): Strategies for handling physical, chemical, and biological noise.

Peter Leopold, BioAnalyte Inc.

Abstract

Biological data measured with modern analytical instruments (mass spec, NMR, UV) is a convolution of useful signal with three kinds of noise: biological, chemical and physical. In this presentation, we will describe how to approach the many degrees of variability in data to ascertain what’s noise and what’s signal. Successful data analysis boils down to the choices of intelligently chosen Bayesian priors coupled with replicate data sets. Several examples, including a mass spectrometry survey of differential protein expression in rat liver, will be discussed.

About the speaker.

Peter Leopold, Ph.D. is a 1985 graduate of Georgetown University and a 1992 graduate of UCSD, where he received a Ph.D. in physics. He was a post-doc in the Harvard Chemistry Dept, a software scientist at Bruker Daltonics, and vice president of ProteiGene Inc. before founding BioAnalyte in Portland in 2001. He has been a Portland resident since 2000. Leopold is on the steering committee of the BioSIG and Software Testing groups at TechMaine.

For more information, see http://www.techmaine.com/biosig-01-26-09

Posted in Uncategorized | No Comments »

“Free” Custom Software for Small Instrument Vendors

February 7th, 2009

Suppose you make scientific instruments that add value to other instruments.

Suppose your instruments rely on other instruments to collect the data. In mass spectrometry, that means you make a source, detector, or robot for another platform company’s analyzer. You’ll quickly confront some grim realities:

  • Your precious raw data are created in the platform company’s proprietary format.
  • Your competitive advantage is now tied up with the platform company’s decision making. Will they give you access to your own data? Or won’t they?
  • And no surprise, the platform company has no burning interest to support your application in its software.

How can you reach your market when you can’t even read — let alone process — your own data?

But it gets worse. If your instrumentation retrofits to a half dozen existing platforms, then you are going to solve this problem for every platform!

The problem is clear: your data is jailed in their format. What are you going to do?

Here are some candidate solutions.

Export the data and build your own tools.

Ask your customers to use the platform company’s software to export the data to some vendor neutral format like XML or ASCII. This is the most common, but also the most annoying to customers. It solves your problem, not theirs.

Advantages

  • Your team can develop code around a standard representation of the data.

Disadvantages

  • Exporting to human readable text invites fraud and bypasses existing checksum protections often available in the proprietary binary form.
  • Exporting makes two “original” copies of the data – one truly original binary and one pseudo-original ASCII/XML — creating ambiguity over process order.
  • Exporting is resource intensive. It requires CPU cycles to parse and write, and then to parse and read again, and it more than doubles disk space use.
  • Exporting is often not automatable. Unless the platform company has an export-to-text function in its scriptable API, your customers will have to export manually.
  • Exporting is not complete. Only the data that the platform company thinks you need will be exported.
  • Even if you export, you’ll still have to hire a team to parse the data and build your application.

License the platform company’s SDK and build your own tools.

Advantages

  • You will eliminate the export requirement and read data faster than by parsing ASCII or XML.

Disadvantages

  • Platform company’s API may not have the functions you really want/need, so only part of the data will be available.
  • The license agreement may not suit your business needs.
  • The API may not suit your development environment. Often, SDKs are dll-based and so are available for Windows-only processing.
  • The API may not be up-to-date with current platform instrumentation. Since the API is not part of the platform company’s mission, it has little incentive to keep it current.
  • The API may be intentionally crippled “for safety.” Unbuffered reads/writes may never break, but they won’t be very fast, either. You should not be surprised if your tools under-perform the platform vendor’s own tools.
  • You still have to hire a software group.

Wait for your application to be supported by the company that controls the format.

Advantages

Zero cost; zero effort.

Disadvantages

  • Forever-to-market
  • Little control over design
  • Vanishing competitive advantage.
  • This is no way to run your business.

There is another way

Describe your application to BioAnalyte. Provide some use cases (benefits) and some data. Also, name a few customers who would be willing to pay ~$2K for a first copy of the software. We’ll do this rest.

We (or you) sell to your customers you (or we) receive a commission. As the applications evolve, we follow you, and all of the funding comes from your customers.

  • We read raw vendor from all of the platform instruments companies.
  • We use our own readers, so we can port them to a web environment, or MacIntosh, or Linux.
  • We meet your specification to a T.
  • No up front cost to you, and no software group salaries on your books.
  • Steady residual revenue source over time as sales increase.
  • And we are as flexible as the market place itself.

Our software is free for you. It is customized for your and your customers’ needs. And it helps you get to the marketplace in a hurry.

Drop us a line. msredux@bioanalyte.com.

Tags: , ,
Posted in Instrumentation | No Comments »

Strip Chart Retro

February 7th, 2009

Remember when the first/best result of an experiment was the data that printed while the experiment ran?

You may have to think back 40 years. If you can’t, then here’s a once-upon-a-time tale from the lab:

It is 1969, and while the world tumbles to the end of the turbulent 60′s, you are safe in your lab worried about vacuum leaks, voltage spikes, flaky power supplies, and whether or not arcing will ruin your experiment, if not also your source.

Perhaps you also are worried about your strip chart recorder. Is the ink fresh? Is the stylus aligned? Is the paper rolling freely? (Is it on?) When you start your experiment, you’ll want to see the data acquired live, and breaking vacuum to reset the source because of a paper jam is not, well, groovy.

But there is one thing you can count on: a nice paper record of your experiment if it all works out OK. That little piece of paper goes into the dated, page-numbered lab notebook. (Too bad that some fraction of the time you have to cut out the peaks to weigh the paper for “semi-quantitative analysis.” ) In fact, that little paper record is always the first (and usually the best) tangible embodiment of your work and career.

Fast-forward 40 years. You live in a world of robotic sample handling, high-throughput analysis, and web-enabled everything. Your newest instrumentation resolves the tiniest features from the smallest amounts of your most complicated samples.  Your GB/hr data acquisition streams through a digital signal processor before it bounces off your hard drive and onto a distributed data sharing network. Dedicated web servers present complex reports on any web client in the network. To satisfy the regulatory authorities, your electronic data are sealed under a 21 CFR Part 11 compliant process.

But really all you wanted was a piece of paper for the lab notebook.

We at BioAnalyte seek to reintroduce the paper trail back into the laboratory. That’s right! Your multi-million dollar high throughput lab will once again experience the charm of a piece of paper that “stickies” right into your lab notebook.

Thermal sticky printing for advanced mass spectrometry is a throw-back to the era of strip chart recorders — all of the utility, none of the cut-and-weigh.

If this sounds nostalgic, then all I can say is “groovy.” But does it sound useful? How would you use? Send us some encouraging words to msredux@bioanalyte.com and become part of the beta test team.

As we like to say, “Your data. Your choice.”

Give us your opinion on a linkedin poll. Vote here.

Tags: , ,
Posted in Thermal sticky printing | No Comments »