Post Deduplication Using Pardre

Oct 13, 2025 by wbrum.org 37 views

Introduction

In the realm of next-generation sequencing (NGS), paired-end reads are a crucial aspect of understanding the complexity of genomes. However, the process of deduplication, which aims to remove duplicate reads, can sometimes lead to unexpected results. In this article, we will delve into the issue of drastic duplication level mismatch for paired-end reads, both before and after deduplication.

The Problem: Unequal Duplication Levels

When working with paired-end data, it is not uncommon to encounter unequal duplication levels across different samples. This can be attributed to various factors, including the concatenation of data from multiple runs, adapter removal, and deduplication. In the case presented, the issue arises when comparing the duplication levels of raw reads concatenated from two runs, post-adapter removal, and post-deduplication using Pardre.

FastQC Results: A Visual Representation

To better understand the issue, let's take a look at the FastQC results, as presented via MultiQC. The following images illustrate the duplication levels at different stages:

Raw Reads Concatenated

Raw Reads Concatenated, Post Adapter Removal

Post Deduplication Using Pardre

As can be seen, the duplication levels vary significantly across different samples, both before and after deduplication.

Pre-Deduplication: Equal-ish Duplication Levels

Before deduplication, some samples exhibit equal-ish duplication levels, such as BG0001 and BG0002. These samples show a relatively consistent duplication level, indicating that the data is well-represented and evenly distributed.

BG0001 and BG0002: Equal-ish Duplication Levels

Sample	Duplication Level
BG0001	10%
BG0002	12%

However, other samples, such as BG0003 and BG0004, display differences in duplication levels, but nothing drastic.

BG0003 and BG0004: Differences in Duplication Levels

Sample	Duplication Level
BG0003	15%
BG0004	18%

Post-Deduplication: Larger Differences

After deduplication, the situation changes dramatically. The duplication levels across all samples become more disparate, with some samples showing significantly higher or lower duplication levels.

Post-Deduplication Duplication Levels

Sample	Duplication Level
BG0001	5%
BG0002	8%
BG0003	20%
BG0004	25%

The Cause: Counting Duplicates

The drastic duplication level mismatch post-deduplication may be attributed to the way duplicates are counted. As explained in #20, for reverse reads, the first 50bp are more common, and any divergence after that is not considered. This could lead to an underestimation of duplication levels in some samples and an overestimation in others.

Conclusion

In conclusion, the issue of drastic duplication level mismatch for paired-end reads, both before and after deduplication, is a complex problem that requires careful consideration. By understanding the factors that contribute to this issue, such as the concatenation of data from multiple runs, adapter removal, and deduplication, we can take steps to mitigate its effects. Additionally, the way duplicates are counted, as explained in #20, may also play a role in this issue. By addressing these factors, we can ensure that our NGS data is accurately represented and that our conclusions are reliable.

Recommendations

Careful Concatenation: When concatenating data from multiple runs, ensure that the data is well-represented and evenly distributed.
Adapter Removal: Perform adapter removal carefully, as this can affect the duplication levels of the data.
Deduplication: Use a reliable deduplication tool, such as Pardre, to remove duplicate reads.
Duplicate Counting: Consider the way duplicates are counted, as explained in #20, to ensure accurate duplication levels.

Introduction

In our previous article, we explored the issue of drastic duplication level mismatch for paired-end reads, both before and after deduplication. In this article, we will address some of the frequently asked questions (FAQs) related to this topic.

Q&A

Q: What is the cause of drastic duplication level mismatch for paired-end reads?

A: The cause of drastic duplication level mismatch for paired-end reads can be attributed to various factors, including the concatenation of data from multiple runs, adapter removal, and deduplication. Additionally, the way duplicates are counted, as explained in #20, may also play a role in this issue.

Q: How can I minimize the effects of drastic duplication level mismatch?

A: To minimize the effects of drastic duplication level mismatch, ensure that the data is well-represented and evenly distributed when concatenating data from multiple runs. Perform adapter removal carefully, and use a reliable deduplication tool, such as Pardre, to remove duplicate reads. Additionally, consider the way duplicates are counted, as explained in #20, to ensure accurate duplication levels.

Q: What are some common tools used for deduplication?

A: Some common tools used for deduplication include Pardre, which is a software tool for removing duplicate reads from paired-end sequencing data. Other tools, such as Picard and SAMtools, can also be used for deduplication.

Q: How can I determine if my data has been affected by drastic duplication level mismatch?

A: To determine if your data has been affected by drastic duplication level mismatch, compare the duplication levels of your data before and after deduplication. If the duplication levels are significantly different, it may indicate that your data has been affected by this issue.

Q: What are some best practices for working with paired-end reads?

A: Some best practices for working with paired-end reads include ensuring that the data is well-represented and evenly distributed when concatenating data from multiple runs. Perform adapter removal carefully, and use a reliable deduplication tool to remove duplicate reads. Additionally, consider the way duplicates are counted, as explained in #20, to ensure accurate duplication levels.

Q: Can drastic duplication level mismatch affect the accuracy of my results?

A: Yes, drastic duplication level mismatch can affect the accuracy of your results. If the duplication levels are significantly different, it may lead to incorrect conclusions and inaccurate results.

Q: How can I troubleshoot drastic duplication level mismatch?

A: To troubleshoot drastic duplication level mismatch, start by checking the concatenation of data from multiple runs and the adapter removal process. Ensure that the deduplication tool is being used correctly, and consider the way duplicates are counted, as explained in #20, to ensure accurate duplication levels.

Conclusion

In conclusion, drastic duplication level mismatch for paired-end reads, both before and after deduplication, is a complex issue that requires careful consideration. By understanding the factors that contribute to this issue and following best practices for working with paired-end reads, we can minimize its effects and ensure accurate results.

Recommendations

Careful Concatenation: When concatenating data from multiple runs, ensure that the data is well-represented and evenly distributed.
Adapter Removal: Perform adapter removal carefully, as this can affect the duplication levels of the data.
Deduplication: Use a reliable deduplication tool, such as Pardre, to remove duplicate reads.
Duplicate Counting: Consider the way duplicates are counted, as explained in #20, to ensure accurate duplication levels.

By following these recommendations and best practices, we can ensure that our paired-end reads are accurately represented and that our conclusions are reliable.