This paper presents a high-speed RNA microextractor for the direct isolation of RNA from peripheral blood lysate using magnetic oligo-dT beads. The extraction is achieved through lateral magnetophoresis, generated by a ferromagnetic wire array inlaid on a glass substrate. This RNA microextractor separated more than 80% of magnetic beads with a flow rate up to 20 ml h(-1), and the overall extraction procedure was completed within 1 min. The absorbance ratio of RNA to protein (A(260)/A(280)) was >1.7, indicating that the extraction technology yielded nearly pure RNA. The feasibility of this technique was evaluated further for its applicability to reverse transcription polymerase chain reaction (RT-PCR) procedures by performing cDNA synthesis and PCR. The analysis verified that the RNA microextractor is a practical method for easy, rapid, and high-precision RT-PCR using minimal reagent volumes without requiring highly trained personnel. In addition, it can be readily incorporated into genetic analysis procedures for realizing automated on-chip genetic platforms in a micro format.

We present an OpenGL ES implementation which utilizes the existing OpenGL library, aiming to support various embedded systems in current consumer markets. Nowadays OpenGL ES becomes an improved version of OpenGL for embedded systems through introducing new features including the fixed-point numeric type. We optimized arithmetic operations on its specific data types such as fixed-point numbers, and achieved totally new optimized implementations of newly introduced features. Efficient ways of parameter conversions between our OpenGL ES implementation and underlying OpenGL are also accomplished even with strictly obeying the standard specifications. Our final implementation result of OpenGL ES 1.1 library completely provides more than 200 API functions in the standard specification and satisfies all the conformance tests. From the efficiency point of view, we compared execution speeds of real world applications to existing commercial implementations to finally show at most 33.147 times speed-ups, which is fastest among the same category implementations.

No abstract