SEQanswers (
-   Bioinformatics (
-   -   QC'ing NGS data (

bw. 11-04-2013 10:39 PM

QC'ing NGS data
Hi all,
What tools are people using for QC'ing the initial stages of NGS data analysis?

I'm thinking tools that run on
- unaligned .fastq
- aligned .bam
- single vs. paired end
- exome seq vs. RNAseq
- possibly whole-genome, chipseq

I'm currently using FastQC and picard - but have been thinking of ways to improve on this (which led to the feature list below). I would appreciate tool recommendations and/or suggestions for additional features you like or don't like in your QC tools.


Ps. Desired features:

* FastQC-like
- user-friendly web interface
- very easy to download, install, run

* Picard-like
- handles both single and paired-end data, aligned and unaligned
- outputs data tables that are easy to read into R, etc.

* Extensibility
- A simple way to add custom graphs and stats would be nice. Although picard and FastQC are both open source and modular, they are written in Java which adds complexity and reduces the potential developer base relative to a language like python (or a toolkit that's not language-specific).
- An open development model (free/open-source) with more than one active contributor (but strong focus on retaining simplicity and not trying to solve all problems).
- That being said, ideally the tool would accumulate knowledge about typical problems/failures in common protocols, and where possible, give suggestions on causes rather than just report graphs/stats/warnings and expect users to know what to do. So where possible, aimed at NGS newbies.

* Holistic
- at the risk of added complexity, the tool could simultaneously take .fastq and one or more processed bam files from a single sample, and then show both unaliged and aligned stats in one interface, as well as how the stats change as a result of the different processing steps in the NGS pipeline.

gringer 11-05-2013 01:22 AM

Sounds like you want a one-size-fits-most solution. Galaxy comes close:

bw. 11-05-2013 04:11 AM

Thanks gringer - I took a look at that Galaxy instance, but didn't see much in the way of QC (other than FastQC).

mgogol 11-05-2013 06:14 AM

I think FastQC and picard are sort of "enough." But I know there are a few in R. Haven't used them.

All times are GMT -8. The time now is 08:58 PM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.