3.3 Features for improving sequencing results

Based on practical experience, we greatly appreciate the automated comparison of observed adapter sequence content and sample sheet. Unexpected sequence in either set is an indication for possible errors. DF Server provides fine-grained control to acknowledge and suppress inconsistency warnings (after either fixing errors or accepting errors and then excluding corresponding data). Furthermore, common artifacts such as PhiX sequence are automatically recognized and show up as information rather than warnings or errors. Figure 2 shows an example.

When adding the sample sheet (not shown), the operator made a small mistake. The adapter P37 is given twice for the same lane in the sample sheet while the adapter sequence ‘AAGACCGT’ occurs in the raw BCLs but not in the sample sheet. This information can then be used for debugging sample sheet information. This is highlighted in the sample sheet (a) and the display of the adapters read from the raw BCL data (b)

The DF Client and Demux components are available from Bioconda as Conda packages and Docker images, thus allowing for future proof installations and creating reproducible workflows. By offering REST APIs and two useful client applications, DF greatly supports sequencing and demultiplexing operators in automating their work. Further automation can be added later as the APIs are open. Automated quality control using FastQC and aggregation using MultiQC also allows users to spot problems earlier (together with the sample sheet adapter checks described earlier). In our experience this allows for the early detection of many common issues. For example, from time to time, it occurs that the same adapter was used for two different libraries in the same lane. This error might be hard to spot on paper or in spreadsheets but applications such as DF Server can easily detect and report such problems similar to the example shown in Figure 2.