The Sequence Read Archive (SRA) is a central repository for raw sequencing data, but its reuse depends on accurate experimental metadata. In 2015, Alnasir and Shanahan measured how often library construction protocols were annotated. Now, Alnasir has repeated that measurement on a September 2026 NCBI metadata snapshot, using the same SQL keyword definitions. The new study, posted on arXiv, covers 41,593,302 experiments, up from 419,620 in the earlier analysis.
The proportion of experiments whose library_construction_protocol field matches all three historical keyword groups—fragmentation, adapter ligation, and enrichment—fell from 4.06% to 1.42%. The protocol field is NULL in 67.65% of contemporary records. When the search also includes design_description, detection rises to 1,197,367 experiments (2.88%), with most additional matches coming from records where the protocol field is empty.
Coverage varies by sequencing strategy, from 0.26% for amplicon to 7.69% for RNA-Seq. The author validated the approach on 150 sampled experiment records against source XML and found agreement in all historical keyword classifications. The findings indicate that the original criteria identify a smaller fraction of experiments in the designated field, and that single-field searches miss preparation details. The paper argues for more consistent protocol annotation and for distinguishing keyword occurrence from the adequacy of experimental documentation.