應用於微流道生物晶片之多反應物樣本製備技術

Abstract

原始的反應試劑或生物樣本藉由一連串的混合及稀釋，來達到適合生化反應進行之特定濃度的液體，該特定濃度稱為目標濃度(target concentration)，而這整個步驟稱為樣本製備程序( sample preparation)，它是各種生化反應中必要的步驟之一。目前大部分提出的樣本製備技術皆應用在支援[1:1]混合模型的數位微流體生物晶片。在多反應物樣本製備程序中，總製備成本為各別原始的反應試劑使用量乘以其試劑成本的總和。CoDOS即為針對總製備成本最小化所發展出適用於數位微流體生物晶片上的多反應物樣本製備演算法，其實驗成果顯著地降低了總製備成本。微流道生物晶片上的混合器可以根據需求劃分成多個相同大小的區段，因此微流道生物晶片支援多種混合模型。越多種的混合模型，代表有越高的可能性可以降低總製備成本。因此本論文將提出針對總製備成本最小化所發展出適用於微流道生物晶片上的多反應物樣本製備演算法。此演算法將會根據混合器提供的多樣混合模型，探索所有可能的混合組合。相較於CoDOS，當混合器劃分成八等分且反應物數量為三和七時，此演算法皆可以減少約70%的總製備成本。

Sample preparation, which is essential to many biochemical reactions, is a process that raw reactants follow the proper mixing sequence to achieve a mixture with a specified concentration value, called target concentration. In recent years, many studies have addressed sample preparation problem on digital microfluidic biochips (DMFBs), and these studies only use the [1:1] mixing model. In the many-reactant sample preparation problem, the cost of a reactant in a dilution process is its usage multiplied by its weight, and the total dilution cost is the summation of every reactant’s cost. CoDOS is the first algorithm aiming at cost minimization in many-reactant single-target sample preparation on DMFBs, and the experimental results show that the improvement in the total dilution cost is significant. A Mixer-N, which is a circular channel divided into N equal-sized segments on flow-based microfluidic biochips (FMFBs), can provide various mixing models. The more mixing models are used in a dilution process, the higher chances to reduce the total dilution cost. Thus, we propose a volume-oriented pattern exploration (VOPE) algorithm, aiming at cost minimization for many-reactant single-target sample preparation problem on FMFBs. Compared with CoDOS, VOPE can reduce the total dilution cost by 70% if the mixer-8 is used and the number of reactants is set to 3 and 7.