Compound

all-cis-docosa-4,7,10,13,16,19-hexaenoic acid

DHA, the specific all-cis isomer present in fish oil. PINNED TO CORRECT AN AUTO-RESOLUTION: the resolver exact-matched CHEBI:36005 'docosahexaenoic acid', which is the generic PARENT CLASS covering any isomer. Left alone, DHA would have been a chemical class while EPA was a molecule — two omega-3 nodes at different levels of specificity. An exact label match can still be the wrong entity. SIX BOUNDARIES. (1) DHA CO-ADMINISTERED WITH EPA IS NOT A DHA TRIAL, and this is the mirror of the boundary on CHEBI:28364. Fish oil and nearly every omega-3 product deliver both, and a pooled result cannot separate their contributions, so it belongs to neither compound alone. WHAT MAKES IT A COMBINATION IS EPA BEING PRESENT ONLY WHERE THE DHA IS. A trial giving the SAME EPA in both arms has EPA as balanced background and its randomised contrast isolates DHA — extract it and record the background in population. So does a DHA-only arm against placebo, and a DHA arm where EPA is the comparator. A trial that doses EPA rather than DHA is not homeless: EPA is CHEBI:28364, so decline subject_species_mismatch and name it in dosed_subject. (2) MOST OF THIS LITERATURE DOSES ONE PERSON AND MEASURES ANOTHER, AND THAT IS RECORDED RATHER THAN DECLINED. Twenty-six of 75 corpus papers give DHA to a pregnant woman, a nursing mother or a preterm neonate and report an outcome in the child — IQ at five years, bronchopulmonary dysplasia, behaviour, growth, ductus arteriosus. These are randomised and they are evidence. They are also the easiest claim in the graph to read wrongly, because the edge says DHA does something and the reader assumes it does that to whoever takes it. `population` MUST NAME BOTH PARTIES IN ORDER — who was dosed, then in whom the outcome was measured — and the dose in dose_regimen is the dose the DOSED party received, never a figure back-calculated for the child. Where a paper reports the same endpoint in mother and infant, they are two claims with two populations, not one. Where the design is a follow-up of a trial randomised years earlier, say so: the randomisation is sound and the interval is long. (3) ARACHIDONIC ACID IS THE OTHER HALF OF THE PRETERM PRODUCTS AND HAS NO NODE. Seven papers give DHA with arachidonic acid to very preterm infants. AA having no node changes nothing about the test, which is the comparator and not the co-agent: DHA plus AA against no supplement is a combination exposure — decline it and name AA in co_exposure — while DHA plus AA against the SAME AA alone is balanced background and isolates DHA. The same reading applies to the trials pairing DHA with piracetam, folate, medium-chain triglycerides or vitamin E. (4) ALGAL OIL IS A PREPARATION OF THIS COMPOUND, NOT A DIFFERENT ONE. Microalgal DHA is the vegetarian source and is chemically the same molecule; so are the triglyceride, ethyl ester and microencapsulated forms. All are this node, with the form written into `preparation` and the wording of the paper kept. Splitting them would divide one literature across nodes that tie on every paper, and discarding the form would hide a bioavailability dispute inside the graph instead of letting it be read off it. This mirrors the ester-form boundary on EPA. (5) THE OMEGA-3 INDEX HAS NO NODE, AND THAT IS WHY THIS SUBJECT DIFFERS FROM VITAMIN D. Where a trial gives DHA and measures the rise in the Omega-3 Index or in red-cell DHA, the finding is real but the graph has nowhere to put it: no outcome node holds either. That is no_outcome_node, and it IS a node gap a curator could close. It is NOT subject_is_the_outcome, which is the different case where DHA is the readout of something else. Where circulating DHA is the EXPOSURE in an observational study, the exposure is a biomarker of intake AND of metabolism rather than of an assigned dose — extract it if the design supports it, at the tier the DESIGN carries, and say in extraction_note that nothing was administered. (6) THE OTHER OMEGA-3S ARE NOT THIS NODE. Alpha-linolenic acid, docosapentaenoic acid (CHEBI:77224) and the oxidised and esterified DHA derivatives are different compounds. So is CHEBI:36005, the parent class this node was pinned away from: a paper about 'docosahexaenoic acid' generically is still about this molecule in practice, but a paper about a different isomer is not.