This is a small-scale study of the effects on microbes in the human gut of adding high-fiber snack foods to usual diet. The snack foods are being provided by Mondelez International, Inc. Eight pairs of dizygotic twins, discordant or concordant for obesity, will participate in the study. They will complete 3 successive experiments in which their regular diet is supplemented by consumption, over a 6-week period, of high-fiber snack foods. Each experiment will involve (i) two weeks consuming regular diet (fecal sample collected weekly); (ii) one week consuming one snack food per day (approximately 7 g fiber, collection of all fecal samples), (iii) one week consuming two snack foods per day (approximately 14 g fiber, collection of all fecal samples), (iv) four weeks consuming three snack foods per day (approximately 20 g fiber, collection of all fecal samples); and (v) two weeks consuming only regular diet (weekly collection of fecal sample). For each experiment, participants will also provide a fasting blood sample during the free diet phase at the end of the week just before initiation of snack consumption, and a fasting blood sample at the end of the last week of snack consumption. Participants will also collect one first morning urine sample weekly throughout the study. During week 2, participants will also collect first morning urine samples for 3 consecutive days. Similarly, they will collect first morning urine samples for the last 3 consecutive days of the last week of snack consumption. The study will test the effects of the different fiber-rich snacks on the composition and metabolic properties of the gut microbial community in lean and obese subjects.

Inclusion Criteria:

* member of female dizygotic twin pair discordant or concordant for obesity

Exclusion Criteria:

• pregnant or trying to get pregnant
• inflammatory bowel disease
• gastrointestinal cancer
• hepatitis
• HIV
• renal failure
• food allergies

Individual data will be shared after completion of the study, to the extent that this can be accomplished without violating the confidentiality of participants (some of whom would be easily identifiable from their clinical metadata).