BackgroundThe microbiome is thought to play a role in the development of gastric cancer (GC). Several studies have put forward putatively carcinogenic species in addition to Helicobacter pylori, but are not in perfect alignment, possibly due to variable parameters in the experiments, including downstream processing. Meta-analyses have not been published so far, so there is lack in clinical guidance beyond H. pylori eradication therapy.MethodsHere, we analysed gastric mucosa samples from nine public data sets, including GC samples. Using both unsupervised and supervised learning, we defined fine grain bacterial networks of gastric mucosa and identified species associated with tumor status of samples.ResultsWe found anatomic locations and cohort regions among the possible factors leading to the observation of study specific gastric microbiomes. Despite this variability, the periodontal species Fusobacterium nucleatum, Parvimonas micra and Peptostreptococcus stomatis were found in association with tumor status in several datasets. The three species were predicted to be in interaction by ecological network analysis and also formed the intersection of tumor associated species between four GC data sets and five colorectal cancer (CRC) data sets we reanalyzed. We formulated a probiotic composition putatively competing with the GC pathogen spectrum, from correlation analysis in a large superset of gut samples (n=17,800) from clinical- and crowd-sourced studies.ImplicationsThe overlapping bacterial pathogen spectrum between two gastrointestinal tumor types, GC and CRC, has implications for etiology, treatment and prevention. In vitro testing results reported in literature suggest H. pylori eradication treatment should be efficient against the GC pathogen spectrum, yet the existence of an upstream periodontal reservoir is of concern. To address this, we propose longer term use of the formulated probiotics composition.