Sequencing the SARS Virus

At 1AM on April 7, 2003, an isolate of the SARS virus arrived at the Michael Smith Genome Sciences Centre. Five days later, the lab published the virus sequence for the first time.
Linux Catches SARS

In the fall of 1999, we received our first DNA sequencer, the MegaBACE 1000 (Figure 6). A sequencer determines the specific base sequence of a DNA sample, though technology currently is limited to determining only 500–800 bases accurately at a time. This read length is much shorter than the size of even the smallest genomes (Tor2/SARS is 30,000 bases in size). Consequently, sequencers simultaneously process 96 samples at a time, and some can be loaded with multiple 96- or 384-well plates.

The MegaBACE is a SCSI device, and the Applied Biosystems (ABI) 3700 and 3730XL sequencers (Figure 6) are controlled through a serial interface and send their data across an Ethernet connection. Although these sequencers acquire large amounts of data in an automated fashion, their software is a point-and-click Windows application. The ABI machines stream their data to a bundled local Oracle database. A UNIX-based control application would revolutionize the deployment of these machines, particularly in large labs. We already have reduced the maintenance complexity of the 3700s by deploying the IBM x330s to replace the original PCs that shipped with the sequencers (Figure 6). Integrating the Windows sequencing platform into a Linux network was the perfect job for smbmount, rsync, Perl and Apache. At the end of each sequence run, the operator triggers a Web-controlled data mirroring process to copy any new data onto the network disks.

After mirroring, the files are first converted from their proprietary format, which encodes the raw signal trace, to the actual bases and their associated quality measure and then are stored in a MySQL database (3.23.55max). Thus far we have collected about 2 million sequencing reads, or about 1TB of raw sequence data.

Figure 3. Representation of our LIMS schema. The Plate table (yellow) references four tables (green) and is referenced by 14 tables (red).

The MySQL Laboratory Information Management System (LIMS) database is central to our sequencing process. Its schema contains 115 tables, 1,171 fields and 195 foreign keys. The database tracks all reagents, equipment, processes and reactions performed in the lab. We circumvent MySQL's lack for native foreign key support by using application logic and a specific field naming convention. Foreign keys are named FKTYPE_TABLE__FIELD, indicating that they point to TABLE_FIELD in the table TABLE. The optional TYPE part of the foreign key name is used to support multiple keys to the same TABLE_FIELD.

Figure 4. Barcodes are printed using networked Zebra printers (left). iPAQs provide a mobile interface into the LIMS (right).

Figure 5. Nearly everything in the lab is barcoded.

Lab technologists interact with the LIMS database using Wi-Fi Compaq iPAQs outfitted with barcode scanners (Figure 4). The iPAQs connect to our internal Apache Web server powering a suite of mod_perl scripts. Objects such as solutions, plates and equipment are barcoded (Figure 5). Barcodes are printed on the networked Zebra S600/96XiIII barcode printers (Figure 4) fed with high-tack labels, which maintain adherence in our –112°F freezers. The barcoding software is written in Perl, uses the ZPL printer language to format the labels and distributes printing using lpr.

Figure 6. Sequencers: 1) MegaBACE 1000, 2) sequencer PC, 3) UPS, 4) sequencer power supply, 5) ABI 3700s, 6) ABI 3730XL and 7) x330 sequencer cluster.

Three generations of sequencers have passed through our lab since the MegaBACE 1000, and we currently operate six ABI 3700s and three ABI 3730XLs (Figure 6). The latest, the ABI 3730XL, is capable of accepting multiple 384-well plates and sequencing 1,152 DNA samples in 24 hours. With each sample yielding up to 700–800 high-quality bases, a single 3730XL produces about 800,000 bases per day.

The Tor2/SARS genome was sequenced using a whole-genome shotgun (WGS) method. In this approach, random sections of the genome are sequenced in a redundant fashion and then assembled together to recover the entire genomic sequence. Given that the size of the pathogen was anticipated to be approximately 30,000 bases, it would take a minimum of 40 reads to span the genome. However, because the reads originate from random regions, more than the minimum number of reads required in order to have enough overlap for a complete assembly. Redundancy also allows for more confidence in determination of the base at each given position in the genome.



Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Silly move. Your loss.

Anonymous's picture

Silly move.

Your loss.