Dear all,
I am dealing with some Ion Proton exome data. The average quality values of each read are very low (all are below 30) and many tests in FastQC report were failed. I've found some information about Ion proton quality values here. I was expecting a good amount of data to pass a filter of Q15 (and also Q10) but I didn't get much ( <1% out of >2.5m reads). I've used prinseq to filter out low quality reads (and also removed reads below the length of 70). I was stuck with following questions.
- Is it a problem with the machine or our data is contaminated during library preparation?
- Is it appropriate to proceed to downstream analysis by trimming the low quality bases of each read at the both ends?
- If the data is of very low quality what should I do with the data (Is it waste of time to go for further analysis?)
I would like to know the views of people who dealt with Ion Proton data previously and I can also provide if any details required.
1 answer
You could use BBMap to map some samples (without quality trimming the reads) to your reference genome and then check its output statistics. BBMap has several options for output statistics, the most interesting for you would probably be "idhist", "ehist" and "indelhist" - check its README file or call bbmap.sh without parameters to see all options.
Log in to answer this question.
-
High quality genome assembly based on poor data with reference genome
written by Jason •
-
Picking parameters for QC of long-read data?
written by Gio •
-
Good tool/best practice for filtering out homopolymeric reads?
written by Sean •
-
Filtering Read Quality - Cant remove all reads that have a single base below a threshold
written by tbb21 •
-
Trouble with transcriptome data
written by gayachit
-
Ion Torrent Proton RNASeq reads alignment
written by mkulecka
-
Can Q10 be better than Q30
written by skbrimer
-
About Ion Proton Data
written by venu
-
SNP calling pipeline - testing different programs
written by Eric Normandeau
-
calling somatic mutations on ion proton data
written by CHANG •
Do u have box plot of per base quality ?
Here is how it looks
This does not look like really bad, how are you processing for QC ? what command are you using to filter ?
After checking the FastQC, I've filtered using
prinseqgivingmin_lenvalue70andmin_qual_score 25. Instead of removing all reads which are of low quality, I trimmed low quality bases at the 3' and 5'. Then FastQC report is a little better than the previous.