Computational Systems Biology Methods and Protocols.7z

has some drawbacks. The first drawback is that local trimming only

uses the quality information for trimming, but it cannot utilize the

global statistical information to discover the abnormal cycles. The

second drawback is local trimming results in unaligned trimming,

which means duplicated reads may be trimmed differently and

consequently causes some deduplication tools like Picard to fail.

Most of these deduplication tools detect duplications only by clus-

tering reads with identical mapping positions. In contrast, AfterQC

implements a global trimming strategy, i.e., it trims all the reads in

the same manner. An algorithm is used to determine how many

cycles to trim in the front and tail, which is based on the segmenta-

tion of the per-cycle base content curves and base quality curves.

A major advantage of AfterQC is the overlapping analysis. Let

Tdenote the length of a sequenced DNA template, andSdenote

the length of paired-end sequencing length, then the pair of reads

will totally overlap ifTS, will overlap with a length of 2ST,if

S<T< 2 S, and will not overlap if 2ST. AfterQC checks how

does each pair of reads overlap based on edit distance optimization.

For a pair of readsR1 andR2, letObe the offset, and we placeR 2

underR1, then we will have vertically aligned subsequencesR 1 o

andR 2 o, and we can calculate their edit distanceed.(R 1 o,R 2 o).

The method optimizes offsetOto obtain the minimal edit distance,

Table 2
Feature comparison of FastQC, Trimmomatic, Cutadapt, and AfterQC

FastQC Trimmomatic Cutadapt AfterQC

Quality control Rich

functionality

Few functionality Few

functionality

Rich functionality

Auto trimming None Read by read Read by read Global trimming

Cutting adapter None Single-end/pair-

end

Single-end/pair-

end

Paired-end only

PolyX filtering None None None Supported

Figure plotting Static Static Static Interactive

Overlap analysis None Cutting adapter

only

None Supported with error

correction

Sequence error

profiling

None None None Supported

Bubble detection None None None Supported

Programming

language

Java Java Python Python, C

Speed Fast Fast Fast Fast only for single-end

data

Bioinformatics Analysis for Cell-Free Tumor DNA Sequencing Data 77