It is a phenomenon I call "morning brain."

The sequence read archives (such as the one hosted by NCBI) as a repository for this sequencing data, uses "compression by reference," a highly-efficient way to compress and store a lot of the data. The raw data that comes off these sequencers is often >99% homologous to the reference genome (such as human, etc), so the most efficient way to compress and store this data is only to record what is different between the sequence output and the reference genome.

How is this not just meaning no redundancy for the "same" digital object. Rather than host 1000 copies of the same file, Amazon minimizes redundancy and every cloud user who hosts the same file has access to the same block(s) of data.

how is this any different than the postal service, which charges senders and recipients do not receive anything except what is being sent?

The experiments in mice were performed at 12.5mg/kg, which would be (for the average 65-kg human) a shocking 812.5mg of Triclosan. If your standard amount of handsoap and toothpaste is 2ml that's like brushing your teeth with a 1/3 solution of triclosan and swallowing it.

Like most of the research in PNAS this was not subjected to the high level of peer review expected in most scholarly journals and this paper got through without regard to its relevance and real-world significance.

At a high enough dose, caffeine causes cancer in lab animals. But not at the doses even Slashdotters consume.

While your statement is correct the cancer risk is minimal.

A typical xray is about 1-10 mrem. You get about 5 mrem of radiation dose from just an airplane ride. If a child lives in a concrete apartment building he/she will receive 100 mrem just from the radiation the concrete gives off. Annual occupational exposure limits for radiation workers are 5000 mrem. A deadly dose of radiation is about 50,000 mrem.

For individual research units, the cost of maintaining the processing power and storage space for these types of projects can be cost-prohibitive. Cloud-based options offer distributed computing power and low-cost storage that is often a more economical solution that paying for the equipment in house, especially when genomic projects can come in spurts rather than a continuous stream.

Disclaimer: I work with large amounts of genomic data and use both in-house and cloud-based analysis tools.