Best pre and post alignment / variant calling QC tools for many WGS/RNA samples 2021

Question

[William]

What are your favorite tools in 2021 for pre and post alignment / variant calling QC tools?

Especially if you are dealing with many large whole genome sequenced samples.

My setup is currently:

FASTQ

• FASTQC: https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
• FASTP: https://github.com/OpenGene/fastp

BAM:

• Samtools stats http://www.htslib.org/doc/samtools-stats.html
• QualiMap: http://qualimap.conesalab.org/

VCF

• BCFTOOLS stats http://samtools.github.io/bcftools/bcftools.html#stats

Summary QC report over many samples and FASTQ/BAM/VCF reports

• MultiQC https://multiqc.info/

This does a reasonable job. But a few things could be improved:

• MultiQC works well for few to medium set of samples, but not for 500+ samples. Report becomes difficult to interpret, and difficult so select samples with issues and drill down to these samples with e.g. sequencing issues . All you get is bee-swarm plot without the possibility to find out which samples have strange QC values.
• Samtools stats does not report coverage by it self. You need to calculate it your self by dividing bases mapped by genome size
• Qualimap makes nice reports, but it's (Java) CPU and memory usage is unreasonable, especially for large genomes and many samples
• Qualiamp reports are difficult to summarize over many samples in MultiQC
• BCFTools stats can't output the desired sample stats in 1 pass. 1 pass per sample gives the best stats that can be loaded in MultiQC (taking long with 100GB+ BCF file read many hundreds of times, or first splitting the multi-sample file to single sample BCF)

So I am wondering what other people use, which tools and how do you summarize the QC results over many samples, with stilling being able to drill down to samples with issues.

[GenoMax]

It sounds like your use case is an outlier. Few people have hundreds of samples so the programs you mention above likely work reasonably well for most.

[ATpoint]

How about:

• make 10 MultiQC reports with 50 samples each or so if you really feel like manually inspecting them.
• mosdepth for coverage, I guess nothing beads it in speed and memory usage

[William]

It is possible to increase the multiqc sample number limits for interactive plots and tables. By creating a modified ~/.multiqc_config.yaml in your home dir, see example https://github.com/ewels/MultiQC/blob/44f28ef0726bc65fd965aa99d5a19f7745c749c4/test/config_example.yaml The limit for the interactive table was 500, so I was just above it. After increasing the sample limit to 5000 the interactive table works with 500+ samples, but the interactive plots are very slow, and the entire html is very slow.