Wednesday, June 25, 2008

Finch 3: Getting Information Out of Your Data

Geospiza's tag line "From Sample to Results" represents the importance of capturing information from all steps in the laboratory process. Data volumes are important and lots of time is being spent discussing the overwhelming volumes of data produced by new data collection technologies like Next Gen sequencers. However, the real issue is not how you are going to store the data, rather it is what are you going to do with it? What do your data mean in the context of your experiment?

The Geospiza FinchLab software system supports the entire laboratory and data analysis workflow to convert sample information into results. What this means is that the system provides a complete set of web-based interfaces and an underlying database to enter information about samples and experiments, track sample preparation steps in the laboratory, link the resulting data back to samples, and process the data to get biological information. Previous posts have focused on information entry, laboratory workflows, and data linking. This post will focus on how data are processed to get biological information.

The ultra-high data output of Next Gen sequencers allows us to use DNA sequencing to ask many new kinds of questions about structural and nucleotide variation and measure several indicators of expression and transcription control on a genome-wide scale. The data produced consists of images, signal intensity data, quality information, and DNA sequences and quality values. For each data collection run, the total collection of data and files can be enormous and can require significant computing resources. While all of the data have to be dealt with in some fashion, some of the data have long-term value while other data are only needed in the short term. The final scientific results will often be produced by comparing data sets created from the DNA sequences and their comparison to reference data.

Next Gen data are processed in three phases.

Next Gen data workflows involve three distinct phases of work: 1. Data are collected from control and experimental samples. 2. Sequence data obtained from each sample are aligned to reference sequence data, or data sets to produce aligned data sets 3. Summaries of the alignment information from the aligned data sets are compared to produce scientific understanding. Each phase has a discrete analytical process and we, and others, call these phases primary data analysis, secondary data analysis and tertiary data analysis.

Primary data analysis involves converting image data to sequence data. The sequence data can be in familiar "ACTG" sequence space or less familiar color space (SOLiD) or flow space (454). Primary data analysis is commonly performed by software provided by the data collection instrument vendor and it is the first place where quality assessment about a sequencing run takes place.

Secondary data analysis creates the data sets that will be further used to develop scientific information. This step involves aligning the sequences from the primary data analyses to reference data. Reference data can be complete genomes, subsets of genomic data like expressed genes, or individual chromosomes. Reference data are chosen in an application specific manner and sometimes multiple reference data sets will be used in an iterative fashion.

Secondary data analysis has two objectives. The first is to determine the quality of the DNA library that was sequenced, from a biological and sample perspective. The primary data analysis supplies quality measurements that can used to determine if the instrument ran properly, or whether the density of beads or clusters were at their optimum to deliver the highest number of high quality reads. However, those data do not tell you about the quality of the samples. Answering questions about sample quality, such as did the DNA library contain systematic artifacts such as sequence bias? Were there high numbers of ligated adaptors or incomplete restriction enzyme digests, or any other factors that would interfere with interpreting the data? These kinds of questions are addressed in the secondary data analysis by aligning your reads to the reference data and seeing that your data make sense.

The second objective of secondary data analysis is to prepare the data sets for tertiary analysis where they will be compared in an experimental fashion. This step involves further manipulation of alignments, typically expressed in very large hard to read algorithm specific tables, to produce data tables that can be consumed by additional software. Speaking of algorithms, there is a large and growing list to choose from. Some are general purpose and others are specific to particular applications, we'll comment more on that later.

Tertiary data analysis represents the third phase of the Next Gen workflow. This phase may involve a simple activity like viewing a data set in a tool like a genome browser so that the frequency of tags can be used to identify promoter sites, patterns of variation, or structural differences. In other experiments, like digital gene expression, tertiary analysis can involve comparing different data sets in a similar fashion to microarray experiments. These kinds of analyses are the most complex; expression measurements need to be normalized between data sets and statistical comparisons need to be made to assess differences.

To summarize, the goal of primary and secondary analysis is to produce well-characterized data sets that can be further compared to obtain scientific results. Well-characterized means that the quality is good for both the run and the samples and that any biologically relevant artifacts are identified, limited, and understood. The workflows for these analyses involve many steps, multiple scientific algorithms, and numerous file formats. The choices of algorithms, data files, data file formats, and overall number of steps depend the kinds of experiments and assays being performed. Despite this complexity there are standard ways to work with Next Gen systems to understand what you have before progressing through each phase.

The Geospiza FinchLab system focuses on helping you with both primary and secondary data analysis.

Friday, June 13, 2008

Finch 3, Linking Samples and Data

One of the big challenges with Next Gen sequencing is linking sample information with data. People tell us: "It's a real problem." "We use Excel, but it is hard." "We're losing track."

Do you find it hard to connect sample information with all the different types of data files? If so you should look at FinchLab.

A review:

About a month ago, I started talking about our third version of the Finch platform and introduced the software requirements for running a modern lab. To review, labs today need software systems that allow them to:

1. Set up different interfaces to collect experimental information
2. Assign specific workflows to experiments
3. Track the workflow steps in the laboratory
4. Prepare samples for data collection runs
5. Link data from the runs back to the original samples
6. Process data according to the needs of the experiment

In FinchLab, order forms are used to first enter sample information into the system. They can be created for specific experiments and the samples entered will, most importantly, be linked to the data that are produced. The process is straightforward. Someone working with the lab, a customer or collaborator, selects the appropriate form and fills out the requested information. Later, an individual in the lab reviews the order and, if everything is okay, chooses the "processing" state from a menu. This action "moves" the samples into the lab where the work will be done. When the samples are ready for data collection they are added to an "Instrument run." The instrument run is Finch's way of tracking which samples go in what well of a plate or lane/chamber on a slide. The samples are added to the instrument and data are collected.

The data

Now comes the fun part. If you have a Next Gen system you'll ultimately end up with 1000's of files scattered in multiple directories. The primary organization for the data will be in unix-style directories, which are like Mac or Windows folders. Within the directories you will find a mix of sequence files, quality files, files that contain information about run metrics and possibly images. You'll have to make decisions about what to save for long-term use and what to archive, or delete.

As noted, the instrument software organizes the data by the instrument run. However, a run can have multiple samples, and the samples can be from different experiments. A single sample can be spread over multiple lanes and chambers of a slide. If you are running a core lab, the samples will come from different customers and your customers often belong to different lab groups. And there is the analysis. The programs that operate on the data require specific formats for input files and produce many kinds of output files. Your challenge is to organize the data so that it is easy to find and access in a logical way. So what do you do?

Organizing data the hard way

If you do not have a data management system, you'll need to write down which samples go with which person, group or experiment. That's pretty simple. You can tape a piece of paper on the instrument and write this down, or you can diligently open a file, commonly an Excel spreadsheet, and record the info there. Not too bad, after all there are only a handful of partitions on a slide (2, 8, 16) and you only run the instrument once or twice a week. If you never upgrade your instrument, or never try and push too many samples through, then you're fine. Of course the less you run your instrument the more your data cost and the goal is to get really good at running your instrument, as frequently as possible. Otherwise you look bad at audit time.

Let's look at a scenario where the instrument is being run at maximal throughput. Over the course of a year, data from between 200 and 1000 slide lanes (chambers) may be collected. These data may be associated with 100's or 1000's of samples and belong to a few or many users in one or many lab groups. The relevant sequence files are between a few hundred megabytes to gigabytes in size; they exist in directories with run quality metrics and possibly analysis results. To sort this out you could have committee meetings to determine whether data should be organized by sample, experiment, user, or group, or you could just pick an organization. Once you've decided on your organization you have to set up access. Does everyone get a unix account? Do you set up SAMBA services? Do you put the data on other systems like Macs and PCs? What if people want to share? The decisions and IT details are endless. Regardless, you'll need a battery of scripts to automate moving data around to meet your organizational scheme. Or you could do something easier.

Organizing data the Finch way

One of FinchLab's many strengths is how it organizes Next Gen data. Because the system tracks samples and users, and has group and permissions models, issues related to data access and sharing are simplified. After a run is complete, the system knows which data files go to what samples. It also knows which samples were submitted by each user. Thus data can be maintained in the run directories that were created by the instrument software to simplify file-based organization. When a run is complete in FinchLab a data link is made to the run directory. The data link informs the system which files go with a run. Data processing routines in the system sort the data into sequences, quality metric files, and other data. At this stage data are associated with samples. Once this is done, the lab has easy access to the data via web pages. The lab can also make decisions about access to data and how to analyze the data. These last two features make FinchLab a powerful system for core labs and research groups. With only few clicks your data are organized by run, user, group, and experiment - and you didn't have to think about it.

Thursday, June 5, 2008

Finishing in the Future

"The data sets are astronomical," "the data that needs to be attached to sequences is unbelievable," and "browsing [data] is incomprehensible." These are just three of the many quotes I heard about the challenges associated with DNA sequencing last week at the "Finishing in the Future Meeting" sponsored by the Joint Genome Institute (JGI) and Los Alamos National Laboratory (LANL).


The two and half day conference, focused on finishing genomic sequences, kicked off with a session on metagenomics. Metagenomics is about isolating DNA from environments and sequencing random molecules to "see what's out there." Excitement for metagenomics is being driven by Next Gen sequencing throughput, because so many sequences can be collected relatively inexpensively. A benefit of being able to collect such large data sets is that we can interrogate organisms that can cannot be cultured. The first talk, "Defining the Human Microbiome: Friends or Family," was presented by Bruce Birren from the Broad Institute of MIT & Harvard. In this talk, we learned about the HMP (Human Microbiome Project), a project dedicated to characterizing the microbes that live on our bodies. It is estimated that microbial cells out number our cells by ten to one. It has long been speculated that our microbiomes are involved in our health and sickness and recent studies are confirming these ideas.

Sequencing technologies continue to increase data throughput

The afternoon session opened with presentations from Roche (454), Illumina, and Applied Biosystems on their respective Next Gen sequencing platforms. Each company presented the strengths of their platform and new discoveries that are being made by virtue of having a lot of data. Each company also presented data on improvements designed to produce even more data and road maps for future improvement to produce even more data. As Haley Fiske from Illumina put it, "we're in the middle of an arms race!" Finally, all the companies are working on molecular barcodes, so that multiple samples can be analyzed within an experiment. So, we started with a lot of data from a sample and are going to a lot of data from a lot of samples. That should add some very nice complexity to sample and data tracking.

A unique perspective

Sydney Brenner opened the second day with a talk on "The Unfinished Genome." The thing I like most about a Sydney Brenner talk is how he puts ideas together. In this talk he presented how one could look at existing data and literature to figure things out or make new discoveries. In one example, he speculated on when the genes for eye development may have first appeared. From the physiology of the eye you can use the biochemistry of vision to identify the genes that encode the various proteins involved in the process. These proteins are often involved in other process, but differ slightly. They arise from gene duplication and modification. So, you can look at gene duplications and measure the age of a duplication by looking at neighboring genes. If a duplication event is old, neighboring genes will be unequal distances apart. You can use this information, along with phylogenetic data, to estimate when the events occurred. Of course this kind of study benefits from more sequence data. Sydney encouraged everyone to keep sequencing.

Sydney closed his talk by making a fun analogy where genomics is like astronomy and thus should have been called "genomy." He supported his analogy by noting that astronomy has astro physic and genomics has genetics. Both are quantitative and measure history and evolution. Astronomy also has astrology, the prediction of an individual's future from the stars. Similarly, folks would like to predict an individual's future from their genes and suggested we call this work "Genology," since it has the same kind of scientific foundation as astrology.

Challenges and solutions

The rest of the conference and posters focused on finishing projects. Today the genome centers are making use of all the platforms to generate large data sets and finish projects. A challenge for genomics is lowering finishing costs. The problem being that generating "draft" data has become so inexpensive and fast that finishing has become a signifiant bottleneck. Finishing is needed to produce the high quality referece sequences that will inform our genomic science, so investigarting ways to lower finishing costs is a worthwhile endeavour. Genome centers are approaching this problem by looking at ways to mix data from different technologies such as 454 and Illumina or SOLiD. They are also developing new and mixed software approaches such as combining multiple assembly algorithms to improve alignments. These efforts are being conducted in conjunction with experiments where mixtures of single pass and paired read data sets are tested to determine optimal approaches for closing gaps.

The take home from this meeting is that, over the coming years, a multitude of new approaches and software programs will emerge to enable genome scale science. The current technology providers are aggressively working to increase data throughput, data quality and read length to make their platforms as flexible as possible. New technology providers are making progress on even higher throughput platforms. Computer scientists are working hard on new algorithms and data visualizations to handle the data. Molecular barcodes will allow for greater numbers of samples per data collection event and increase sample tracking complexity.

The bottom line

Individual research groups will continue to have increasing access to "genome center scale" technology. However, the challenges with sample tracking, data management, and data analysis will be daunting. Research groups with interesting problems will be cut off from these technologies unless they have access to cost-effective, robust informatics infrastructures. They will need help setting up their labs, organizing the data, and making use of new and emerging software technologies.

That's where Geospiza can help.