Useful Links

From Bioinformatics Core Wiki

NGS data analysis Protocols, Methods & Tools




How many replicates?
Approaches and benchmarks

Single cell RNA-seq


  • ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res, 2012.
  • deepTools, including plotFingerprint that addresses the question "Did my ChIP-seq work?" by sampling indexed BAM files and plotting a profile of cumulative read coverages for each file.
  • Features that define the best ChIP-seq peak calling algorithms. Briefings in Bioinformatics, 2016. - Benchmarking of peak-calling algorithms.
  • Practical Guidelines for the Comprehensive Analysis of ChIP-seq Data. PLOS Comp Biol, 2013:
    • "For mammalian transcription factors (TFs) and chromatin modifications such as enhancer-associated histone marks, which are typically localized at specific, narrow sites and have on the order of thousands of binding sites, 20 million reads may be adequate (4 million reads for worm and fly TFs)."
    • "Proteins with more binding sites (e.g., RNA Pol II) or broader factors, including most histone marks, will require more reads, up to 60 million for mammalian ChIP-seq."
    • "Importantly, control samples should be sequenced significantly deeper than the ChIP ones in a TF experiment and in experiments involving diffused broad-domain chromatin data. This is to ensure sufficient coverage of a substantial portion of the genome and non-repetitive autosomal DNA regions."
    • "To ensure that the chosen sequencing depth was adequate, a saturation analysis is recommended—the peaks called should be consistent when the next two steps (read mapping and peak calling) are performed on increasing numbers of reads chosen at random from the actual reads. Saturation analysis is built into some peak callers (e.g., SPP, an R package for analysis of ChIP-seq and other functional sequencing data ). If this shows that the number of reads is not adequate, reads from technical replicate experiments can be combined."
    • "To avoid over-sequencing and estimate an optimal sequencing depth, it is important to take into account library complexity." Several tools are available for this purpose: the Preseq package allows users to predict the number of redundant reads from a given sequencing depth and how many will be expected from additional sequencing."


Genome/Transcriptome Assembly

Gene Set Enrichment Analysis (GSEA) and other post-processing analysis

NGS other

  • Nanopore (MinION) de novo bacterial genome sequencing [1]
    • "Many bacterial genomes can be assembled into single contigs if reads longer than 7 kb are available, as these reads span the conserved rRNA operon, which is typically the longest repeat sequence in a bacterial genome.
    • Recent versions of nanopore chemistry (R7.3) coupled with the latest base caller (Metrichor versions 1.9 and later) permit read-level accuracies of 78–85% (refs. 1,8). Although this is slightly lower than accuracies achieved by the latest version of Pacific Biosciences chemistry."
    • Two-dimentional reads from four separate MinION runs using R7.3 chemistry were combined. In total, 22,270 2D reads were used comprising 133.6 Mb of read data, representing ~29× theoretical coverage of the 4.6-Mb E. coli K-12 MG1655 reference genome.
    • Potential overlaps between the reads were detected using the DALIGNER software. Each read and its overlapped reads were used as input to the partial-order alignment (POA) software, which iteratively computes the consensus sequence. The read error-correction software, Nanocorrect, is available at
    • The reads resulting from two rounds of correction were used as input to version 8.2 of the Celera Assembler. This resulted in a highly contiguous assembly with three contigs, the largest being 4.6 Mb long and covering the entire E. coli chromosome.
    • The authors implemented an algorithm that uses the electric current signal to compute an improved consensus sequence for the assembly. That allowed the base-level accuracy improved to 99.5%, comprising 1,202 mismatches (26 per 100 kb) and 17,241 indels of ≥1 base (371 errors per 100 kb). The signal-level consensus software, Nanopolish, is available at
    • The complete pipeline used to generate the assembly, including downloading the input data and required software, is provided as a Makefile on GitHub at Additional scripts used to analyze the assembly are provided in the same repository. An IPython notebook documenting the analysis workflow is also provided.


Data Science


Experimental Design

Statistical Rituals & Statistical Power

Online Resources & Courses

Comparison of two samples

  • The t-test, paired or unpaired, in R >t.test (x,y, paired=TRUE). The t-test provides an exact test for the equality of the means of two normal populations with unknown, but equal, variances. The latter can be checked with F-test, or in R >var.test(x,y).'s_t-test#Paired_samples

Comparison of two microbiome samples

Other topics


Online Courses & Materials

Bioinformatics training providers

Communities & Blogs

Bioinformatics Core Facility @ CRG — 2011-2023