Relative performance of customized and universal probe sets in target enrichment: A case study in subtribe Malinae

Article of: Roman Ufimov, Vojtěch Zeisek, Soňa Píšová, William J. Baker, Tomáš Fér, Marcela van Loo, Christoph Dobeš and Roswitha Schmickl

Introduction: What was know before this publication?

• Universal probes
• more consistent target locus recovery for outgroup taxa
• Customized probes
• higher total number of target loci
• higher locus recovery in the ingroup

​

• Angiosperms353 x custom probes
• similar degree of phylogenetic informativeness

Angiosperms353, Source: https://arborbiosci.com

Introduction: Why was this study necessary?

• Angiosperms353 markers
• Utility has not yet been fully evaluated
• No attempt to optimize it for target enrichment in the data analysis
• Lack of studies on this topic
• One study - comparison of Angiosperms353 x custome probes (Larridon et al., 2020)
• Two studies - comparison of universal probes x custom probes (Kadlec et al., 2017; Chau et al., 2018)

Introduction: What were the aims of study?

​

• Evaluation of the strengths and weaknesses of Angiosperms353
• Discovering phylogenetic utility of Angiosperms353:
• in comparison with Malinae481
• comparison with optimized Angiosperms353

Focused on the genus Crataegus L.

Source: Xiao Du et al. (2019), Frontiers in Plant Science 10:1-12

Methods: Sampling and wet-lab

• The subtribe Malinae (up to 30 genera+10 hybrid) appears to be monophyletic. Level of divergence between the genera is low.
• Selection of 25 species within the Malinae: 13 from Crataegus, 12 from 7 other genera and Prunus tenella (Amygdaleae) as an outgroup.
• The libraries were prepared using the NEBNext Ultra DNA Library Prep protocol and hybridization was done using MyBaits biotinylated RNA baits.
• Target-enriched libraries were mixed with unenriched libraries for better plastome recovery

Methods: Processing nuclear data

• Trimmed and deduplicated reads were analysed with HybPiper (assembly approach) and HybPhyloMaker (full-pipeline; reference-guided mapping approach)
• HybPiper output was aligned with MAFFT and trimmed with custom R script.
• Individual gene trees were inferred with IQ-Tree according to best-fitting model rated by ModelFinder
• Species tree was reconstructed with ASTRAL and support calculated with PhyParts
• Comparison was made between two HybPiper runs and HybPhyloMaker

Methods: Processing plastid data

• HybPhyloMaker was used to map preprocessed reads to the modified plastome reference of M. angustifolia (GenBank). Kindel was then used for consensus calling (min. read depth 2x, majority rule 51 %)
• Number of recovered plastid reads were compared for both probe sets, reads from both sets were then used to infer phylogeny:
• The sequences were aligned using MAFFT, concatenated and partitioned with AMAS, so that each partition included either coding sequences or non-coding sequences
• The plastome tree was inferred using RAxML-NG, using best-fit model for each partition according to ModelTest-NG results

Probe and reference sets

• Two probe sets and four references were used: Angiosperm353 and custom Malinae481
• Probes were designed by:
1. BLASTing Malus domestica “Golden Delicious” mRNAs against M. domestica genome and a two-step filtering.
2. 1280 filtered mRNAs were then BLASTed against Pyrus communis genome to find common mRNAs and filtered with the two-step filtering.
3. Exons and introns were then extracted and separated. Introns underwent yet another filtering
• Two reference sets correspond to probe sets: Angiosperms353 and Malinae481. The other two correspond to finetuned versions of Angiosperm353

Results: Comparison of Angiosperms353 with Malinae481

Both show a decrease in the number of loci, but differently!!!! → Most striking difference in the performance of the probes.

Higher target length in ingroup for Malinae481 and lower for Angiosperms353

Table 3: Alignment characteristics

!

!

Figure 2: Alignment length and proportion of PI sites

​

Results: Bioinformatic optimization of the Angiosperms353 reference for the Malinae (Malinae-optimized)

Malinae-optimized reference performed similarly to Angiosperms353, but nº of loci decreased slower in Malinae-optimized

Similar target lengths for both ingroup and outgroup

Table 3: Alignment characteristics