This dataset comprises the sequence of 44 278 RNA oligonucleotide "baits" (120 bp each) designed to perform whole-genome capture and sequencing of Mycobacterium tuberculosis directly from clinical samples (DNA) using Agilent Technologies’ SureSelect target enrichment system following the Illumina paired-end multiplexed sequencing library protocol. RNA oligonucleotide “baits” were designed to span the ∼4.5 Mb of the M. tuberculosis genome. In brief, the reference genome sequence of the MTBC H37Rv strain (Genbank #AL123456) was in silico fragmented into 120 bp sequences twice, to ensure an overlap of 60 bp between sequences. Due to their rich GC content, which could interfere with DNA capture, all MTBC genes of the PE, PPE and PE-PGRS family were also independently fragmented into 120 bp sequences, in order to increase capture sensitivity. All resulting sequences were BLASTn searched against the Human Genomic + Transcript database to excluded homologous sequences to the human genome. Overall, a total of 42,278 RNA probes were generated and this custom bait library was then uploaded to the SureDesign software (https://earray.chem.agilent.com/suredesign) and synthesized by Agilent Technologies. During synthesis, the 2198 sequences complementary to the PE, PPE and PE-PGRS family were unbalanced 8:1 to potentiate capture. More details can be found in the following publication: - Macedo, R., Isidro, J., Ferreira, R., Pinto, M., Borges, V., Duarte, S., Vieira, L., & Gomes, J. P. (2023). Molecular Capture of Mycobacterium tuberculosis Genomes Directly from Clinical Samples: A Potential Backup Approach for Epidemiological and Drug Susceptibility Inferences. International journal of molecular sciences, 24(3), 2912. https://doi.org/10.3390/ijms24032912